{
 "S1sh3::bgfirst_bg": {
  "input_fingerprint": "d06d1e0092ea4f84",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 과거의 조선인 육혈귀 생존자들이 공포에 질린 표정으로 한 발은 땅을 박차고 다른 발은 앞으로 내뻗은 채 갈대숲 사이를 다급하게 내달리는 찰나.\n\nLOCATION (lock): Among the dense winter reeds of a colonial-era field, along the fugitives' path through the vegetation.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At waist height, the camera tracks parallel to the fleeing group from a lateral position, holding a wide view that keeps terrified faces and opposing running-leg positions legible among the reeds. The survivors spread across the middle band of the frame, each caught between a planted rear foot and an outstretched leading foot, while the lateral motion preserves their forward momentum toward off-screen safety.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 겨울 갈대숲 (The fleeing survivors are moving through it); used as The reeds intermittently cross the foreground and separate individual runners without obscuring the group action.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 과거의 조선인 육혈귀 생존자들이 공포에 질린 표정으로 한 발은 땅을 박차고 다른 발은 앞으로 내뻗은 채 갈대숲 사이를 다급하게 내달리는 찰나.\n\nLOCATION (lock): Among the dense winter reeds of a colonial-era field, along the fugitives' path through the vegetation.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At waist height, the camera tracks parallel to the fleeing group from a lateral position, holding a wide view that keeps terrified faces and opposing running-leg positions legible among the reeds. The survivors spread across the middle band of the frame, each caught between a planted rear foot and an outstretched leading foot, while the lateral motion preserves their forward momentum toward off-screen safety.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 겨울 갈대숲 (The fleeing survivors are moving through it); used as The reeds intermittently cross the foreground and separate individual runners without obscuring the group action.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S1sh3__bgfirst_bg.png",
  "asset_id": "4ede6c81-1631-4dac-8718-1f6c6488ade6",
  "input_asset_ids": [
   "c5247545-53e2-40f4-beaa-f4d12c2ddd28",
   "e1eaa73d-acc3-4d86-9b55-044077e18bd8"
  ]
 },
 "S1sh3": {
  "input_fingerprint": "6736be0f7d05d3c9",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 과거의 조선인 육혈귀 생존자들이 공포에 질린 표정으로 한 발은 땅을 박차고 다른 발은 앞으로 내뻗은 채 갈대숲 사이를 다급하게 내달리는 찰나.\n\nLOCATION (lock): Among the dense winter reeds of a colonial-era field, along the fugitives' path through the vegetation. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 과거의 조선인 육혈귀 생존자 (조선인 남녀노소, 상투와 쪽머리, 굶주린 얼굴, 시대극 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 과거의 조선인 육혈귀 생존자들이 공포에 질린 표정으로 한 발은 땅을 박차고 다른 발은 앞으로 내뻗은 채 갈대숲 사이를 다급하게 내달리는 찰나.\n\nLOCATION (lock): Among the dense winter reeds of a colonial-era field, along the fugitives' path through the vegetation. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 과거의 조선인 육혈귀 생존자 (조선인 남녀노소, 상투와 쪽머리, 굶주린 얼굴, 시대극 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 과거의 조선인 육혈귀 생존자들이 공포에 질린 표정으로 한 발은 땅을 박차고 다른 발은 앞으로 내뻗은 채 갈대숲 사이를 다급하게 내달리는 찰나.\n\nLOCATION (lock): Among the dense winter reeds of a colonial-era field, along the fugitives' path through the vegetation. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 과거의 조선인 육혈귀 생존자 (조선인 남녀노소, 상투와 쪽머리, 굶주린 얼굴, 시대극 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S1sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S1sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 과거의 조선인 육혈귀 생존자: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:916312>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L01B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 과거의 조선인 육혈귀 생존자: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:916312>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "공포에 질린 표정과 다급하게 달리는 자세를 훌륭하게 구현했으며, 레퍼런스의 인물과 배경(열린 문)을 정확히 반영했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "우측 인물의 발이 식물 뿌리처럼 뭉개지는 심각한 신체 묘사 오류가 있으며, 배경의 오두막 문이 닫혀 있어 레퍼런스와 어긋납니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "공포에 질린 표정과 다급하게 달리는 자세를 훌륭하게 구현했으며, 레퍼런스의 인물과 배경(열린 문)을 정확히 반영했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "우측 인물의 발이 식물 뿌리처럼 뭉개지는 심각한 신체 묘사 오류가 있으며, 배경의 오두막 문이 닫혀 있어 레퍼런스와 어긋납니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "'생존자들'이라는 복수 주어를 올바르게 등장시켰으며, '한 발을 박차고 다른 발을 내뻗은' 질주 동작을 측면 구도로 정확히 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "복수 주어(생존자들)임에도 단 한 명만 묘사되었고, 정면 구도 탓에 프롬프트가 요구한 다리 동작이 명확하게 드러나지 않습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "'생존자들'이라는 복수 주어를 올바르게 등장시켰으며, '한 발을 박차고 다른 발을 내뻗은' 질주 동작을 측면 구도로 정확히 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "복수 주어(생존자들)임에도 단 한 명만 묘사되었고, 정면 구도 탓에 프롬프트가 요구한 다리 동작이 명확하게 드러나지 않습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 10,
     "B": 11
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 10,
   "B": 11
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "공포에 질린 표정과 다급하게 달리는 자세를 훌륭하게 구현했으며, 레퍼런스의 인물과 배경(열린 문)을 정확히 반영했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "우측 인물의 발이 식물 뿌리처럼 뭉개지는 심각한 신체 묘사 오류가 있으며, 배경의 오두막 문이 닫혀 있어 레퍼런스와 어긋납니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L01B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 과거의 조선인 육혈귀 생존자: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:916312>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "치마 아래 캐릭터의 왼쪽 다리(화면 우측) 뒤편에 정체불명의 맨발이 하나 더 허공에 떠 있습니다.",
     "fix_en": "Remove the extra bare foot dangling behind the character's left leg.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the extra bare foot dangling behind the character's left leg.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S1sh3__bgfirst_bg.png",
   "bg_asset_id": "4ede6c81-1631-4dac-8718-1f6c6488ade6",
   "bg_record_key": "S1sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S1sh5": {
  "input_fingerprint": "dc106055215bc110",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 일본군 병사들이 갈대숲 뒤편에서 소총을 겨눈 채 화염을 뿜어내는 순간.\n\nLOCATION (lock): Behind the fleeing group in another section of the winter reed field, where soldiers advance and fire through the reeds. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The Koreans already shot down remain motionless among the reeds, with blood staining the area.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 일본군 병사 (일본인 성인 남성, 짧은 검은 머리, 각진 얼굴, 엄격한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 4,
   "A": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "갈대숲에서 사격하는 일본군과 바닥에 쓰러진 조선인들의 묘사가 지문과 이전 장면의 상태 지침을 정확히 따랐습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "인물 구성은 맞으나, 뒤쪽 병사의 소총 총열이 앞쪽 병사의 모자를 관통하는 등 물리적으로 불가능한 오류가 있습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "지문이 명시하지 않은 생존 피난민(서 있는 인물)을 화면 앞쪽에 임의로 배치하여 인물 구성 지침을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S1sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 일본군 병사: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:859073>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "일본군 병사들의 군복 깃에 레퍼런스 이미지에 명확히 보이는 붉은색 계급장(옷깃장)이 모두 누락되었습니다.",
     "fix_en": "Add the red collar tabs with gold insignia to the collars of all visible Japanese soldiers' uniforms, matching the reference image.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "전경에 위치한 병사의 얼굴이 캐릭터 레퍼런스 이미지의 인물(각진 턱, 얇은 콧수염 등)과 일치하지 않습니다.",
     "fix_en": "Modify the foreground soldier's face to exactly match the specific facial features, including the faint mustache and square jaw, of the character reference image.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "우측 바닥에 쓰러진 시신들의 팔다리가 형태학적으로 심하게 엉켜 있으며, 특히 한 시신의 머리가 몸과 분리되거나 비정상적인 각도로 꺾여 있습니다.",
     "fix_en": "Correct the anatomy of the dead bodies on the ground to the right, ensuring their limbs are distinct and heads are properly attached and angled like normal human anatomy.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "전경에 있는 병사의 소총 가죽 멜빵이 총 개머리판 쪽에 연결되지 않고 병사의 팔과 가슴 안으로 융합되어 들어가는 오류가 있습니다.",
     "fix_en": "Redraw the rifle sling on the foreground soldier's weapon so it connects cleanly to the bottom of the rifle stock, rather than merging into his arm and torso.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the red collar tabs with gold insignia to the collars of all visible Japanese soldiers' uniforms, matching the reference image.\n- Modify the foreground soldier's face to exactly match the specific facial features, including the faint mustache and square jaw, of the character reference image.\n- Correct the anatomy of the dead bodies on the ground to the right, ensuring their limbs are distinct and heads are properly attached and angled like normal human anatomy.\n- Redraw the rifle sling on the foreground soldier's weapon so it connects cleanly to the bottom of the rifle stock, rather than merging into his arm and torso.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S1sh3"
  }
 },
 "S1sh7": {
  "input_fingerprint": "b14935e96fb505f0",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 과거의 조선인 육혈귀 생존자들이 낡은 나무 소금창고 문 안쪽을 향해 몸을 던져 허공에 떠 있는 뒷모습.\n\nLOCATION (lock): Directly outside the doorway of a small wooden storehouse standing amid the winter reeds. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The Koreans already killed or incapacitated remain scattered in the reed field behind the fleeing survivors.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 과거의 조선인 육혈귀 생존자 (조선인 남녀노소, 상투와 쪽머리, 굶주린 얼굴, 시대극 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 3,
   "B": 2
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "카메라가 외부에 위치해 문 안쪽으로 향하는 올바른 방향을 연출했으나, 지정된 캐릭터 외형은 누락되었습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "카메라가 내부에 위치해 바깥으로 뛰는 심각한 방향 오류(Hard Violation)가 있으며, 지문에 없는 일본군이 등장합니다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "완전히 복제된 인물 묘사와 지문에 없는 일본군이 등장하는 심각한 규칙 위반(Hard Violation)이 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S1sh5_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 과거의 조선인 육혈귀 생존자: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:916312>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "등장인물들이 레퍼런스 이미지에 제시된 단발머리 여성 캐릭터와 일치하지 않고 상투를 튼 남성들로 그려졌습니다.",
     "fix_en": "Replace the men with the specific woman from the character reference, ensuring her face, short hair, and exact clothing match.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "창고 문 안으로 뛰어드는 인물의 왼쪽 다리 하단과 발이 그려지지 않았습니다.",
     "fix_en": "Draw the missing left foot and straw shoe on the jumping character's left leg.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "프롬프트는 '생존자들이' 허공에 떠 있는 모습을 요구했으나, 한 명을 제외한 나머지 인물들은 땅에 서 있습니다.",
     "fix_en": "Modify the standing characters in the foreground so they are also suspended in mid-air, diving toward the doorway.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "갈대밭에 쓰러진 시체들의 자세와 배치가 이전 샷(레퍼런스)과 일치하지 않습니다.",
     "fix_en": "Replace the dead bodies on the ground to match their exact poses, positions, and arrangement from the previous shot still.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the men with the specific woman from the character reference, ensuring her face, short hair, and exact clothing match.\n- Draw the missing left foot and straw shoe on the jumping character's left leg.\n- Modify the standing characters in the foreground so they are also suspended in mid-air, diving toward the doorway.\n- Replace the dead bodies on the ground to match their exact poses, positions, and arrangement from the previous shot still.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S1sh5"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S2sh3::bgfirst_bg": {
  "input_fingerprint": "952daa7b1f283d60",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 벌컥 열린 창고 문턱에 선 일본군 병사들이 안쪽을 향해 비릿한 웃음을 짓는 찰나.\n\nLOCATION (lock): Just inside the rough wooden storehouse, facing its suddenly opened entrance and daylight beyond.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: On the central interior axis at eye level, the dolly-in ends close enough to make the soldiers' inward-looking smiles readable while retaining the full opened doorway around them. The foremost soldier occupies one side of the middle frame in three-quarter view, with additional soldiers layered behind him at the threshold and the warehouse interior framing the image edges.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: opened warehouse doorway in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 창고 문 (Suddenly opened) — The interior-facing side and open passage are visible from inside the warehouse; used as The opened doorway encloses the invading soldiers and creates a hard boundary between interior and exterior; 덧댄 판자와 각목 (Roughly assembled as part of the temporary warehouse) — Their interior-facing sides border the frame and lead toward the entrance; used as The rough interior structure forms an irregular frame around the doorway.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 벌컥 열린 창고 문턱에 선 일본군 병사들이 안쪽을 향해 비릿한 웃음을 짓는 찰나.\n\nLOCATION (lock): Just inside the rough wooden storehouse, facing its suddenly opened entrance and daylight beyond.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: On the central interior axis at eye level, the dolly-in ends close enough to make the soldiers' inward-looking smiles readable while retaining the full opened doorway around them. The foremost soldier occupies one side of the middle frame in three-quarter view, with additional soldiers layered behind him at the threshold and the warehouse interior framing the image edges.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: opened warehouse doorway in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 창고 문 (Suddenly opened) — The interior-facing side and open passage are visible from inside the warehouse; used as The opened doorway encloses the invading soldiers and creates a hard boundary between interior and exterior; 덧댄 판자와 각목 (Roughly assembled as part of the temporary warehouse) — Their interior-facing sides border the frame and lead toward the entrance; used as The rough interior structure forms an irregular frame around the doorway.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S2sh3__bgfirst_bg.png",
  "asset_id": "f5f91da5-21a7-49c3-898a-f77ae1da09a7",
  "input_asset_ids": [
   "816bdd6e-edbf-4323-b486-e523eb661a84",
   "5bbbaa8d-856c-4efb-b92c-c7ed1b9d9ac2"
  ]
 },
 "S2sh3": {
  "input_fingerprint": "6d7f036ed79c6cc8",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 벌컥 열린 창고 문턱에 선 일본군 병사들이 안쪽을 향해 비릿한 웃음을 짓는 찰나.\n\nLOCATION (lock): Just inside the rough wooden storehouse, facing its suddenly opened entrance and daylight beyond. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 일본군 병사 (일본인 성인 남성, 짧은 검은 머리, 각진 얼굴, 엄격한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 벌컥 열린 창고 문턱에 선 일본군 병사들이 안쪽을 향해 비릿한 웃음을 짓는 찰나.\n\nLOCATION (lock): Just inside the rough wooden storehouse, facing its suddenly opened entrance and daylight beyond. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 일본군 병사 (일본인 성인 남성, 짧은 검은 머리, 각진 얼굴, 엄격한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 벌컥 열린 창고 문턱에 선 일본군 병사들이 안쪽을 향해 비릿한 웃음을 짓는 찰나.\n\nLOCATION (lock): Just inside the rough wooden storehouse, facing its suddenly opened entrance and daylight beyond. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 일본군 병사 (일본인 성인 남성, 짧은 검은 머리, 각진 얼굴, 엄격한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S2sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S2sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 일본군 병사: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:859073>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L02B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 일본군 병사: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:859073>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "화면 좌측에 흰색 화살표 마커가 잘못 유출되어 심각한 규정 위반(Hard Violation)에 해당합니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "창고 문턱에 서서 안쪽을 향해 비릿하게 웃는 일본군 병사들의 모습과 지정된 공간적 배경을 충실하게 구현했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "화면 좌측에 흰색 화살표 마커가 잘못 유출되어 심각한 규정 위반(Hard Violation)에 해당합니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "창고 문턱에 서서 안쪽을 향해 비릿하게 웃는 일본군 병사들의 모습과 지정된 공간적 배경을 충실하게 구현했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시된 장소와 복식을 정확히 반영하고, 창고 안쪽을 향해 비릿하게 웃는 병사들의 표정과 상황을 사실적으로 잘 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "화면 좌측 벽면에 지시용 화살표 마커가 그대로 노출되는 치명적인 규정 위반(Hard Violation)이 발생하여 실격입니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시된 장소와 복식을 정확히 반영하고, 창고 안쪽을 향해 비릿하게 웃는 병사들의 표정과 상황을 사실적으로 잘 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "화면 좌측 벽면에 지시용 화살표 마커가 그대로 노출되는 치명적인 규정 위반(Hard Violation)이 발생하여 실격입니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "화면 좌측에 흰색 화살표 마커가 잘못 유출되어 심각한 규정 위반(Hard Violation)에 해당합니다."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "창고 문턱에 서서 안쪽을 향해 비릿하게 웃는 일본군 병사들의 모습과 지정된 공간적 배경을 충실하게 구현했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L02B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 일본군 병사: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:859073>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "앞쪽 병사가 문턱에 서 있지 않고 창고 안쪽으로 완전히 들어와 있습니다.",
     "fix_en": "Move the front soldier back so he is standing directly on the wooden threshold of the open door.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "뒤쪽 병사의 목 앞쪽 허공에 정체불명의 가로 막대(또는 총검 형태)가 떠 있습니다.",
     "fix_en": "Remove the floating horizontal rod passing in front of the background soldier's neck and cleanly restore his uniform and the doorframe behind it.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "앞쪽 병사가 쥐고 있는 소총의 하단 개머리판 부분이 코트와 섞여 형태가 잘려 있습니다.",
     "fix_en": "Render the bottom wooden stock of the front soldier's rifle clearly and continuously over his coat.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the floating horizontal rod passing in front of the background soldier's neck and cleanly restore his uniform and the doorframe behind it.\n- Render the bottom wooden stock of the front soldier's rifle clearly and continuously over his coat.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S2sh3__bgfirst_bg.png",
   "bg_asset_id": "f5f91da5-21a7-49c3-898a-f77ae1da09a7",
   "bg_record_key": "S2sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S2sh5": {
  "input_fingerprint": "7862e8a0a4e215ef",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 일본군 병사들의 몸이 문밖 어둠 속을 향해 강하게 꺾인 채 허공에 떠 있는 mid-action 자세.\n\nLOCATION (lock): At the inner side of the wooden storehouse doorway, where the soldiers are being dragged out through the entrance. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the rough plank-built warehouse interior, doorway, and daylight entering through it. Exclude the soldiers' earlier smiles and preserve only figures required in the new action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 일본군 병사 (일본인 성인 남성, 짧은 검은 머리, 각진 얼굴, 엄격한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 6,
   "B": 4,
   "A": 2
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "두 병사가 허공에 뜬 모습을 오류 없이 구현했으나, 강하게 꺾인 묘사와 밖으로 끌려가는 방향성은 부족합니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "공중에 뜬 병사가 한 명뿐이며 몸이 전혀 꺾이지 않고 수평으로 떠 있어 제시어 반영도가 낮습니다."
   },
   {
    "label": "A",
    "score": 2,
    "verdict_ko": "두 인물의 다리가 하나로 융합되는 치명적인 해부학적 오류(Hard Violation)가 발생했으며 임의의 검은 형체를 추가했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S2sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 일본군 병사: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:859073>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "병사들이 꼿꼿한 자세로 공중에 떠서 안쪽으로 들어오는 것처럼 보입니다. 문밖으로 끌려가며 몸이 바깥쪽을 향해 강하게 뒤로 꺾인 자세로 수정해야 합니다.",
     "fix_en": "Redraw the soldiers so their bodies are violently arched backward toward the outside, showing them being dragged out through the doorway.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "왼쪽 병사가 든 소총의 앞부분이 오른쪽 병사의 가슴 및 장비와 융합되어 형태가 뭉개져 있습니다.",
     "fix_en": "Redraw the left soldier's rifle so its barrel is distinctly separate and does not melt into the right soldier's body or weapon.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the soldiers so their bodies are violently arched backward toward the outside, showing them being dragged out through the doorway.\n- Redraw the left soldier's rifle so its barrel is distinctly separate and does not melt into the right soldier's body or weapon.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S2sh3"
  }
 },
 "S2sh6": {
  "input_fingerprint": "ecc11916862d19a2",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 굳게 닫힌 나무 문 앞에서 과거의 조선인 육혈귀 생존자들이 충격에 빠져 굳어버린 찰나.\n\nLOCATION (lock): Inside the dim wooden storehouse, immediately before its firmly shut plank door. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the rough timber walls, closed-door position, and dim daytime interior lighting. Exclude the soldiers at the threshold and the earlier open-door state.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 과거의 조선인 육혈귀 생존자 (조선인 남녀노소, 상투와 쪽머리, 굶주린 얼굴, 시대극 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 5,
   "C": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "굳게 닫힌 문 앞에서 충격에 빠진 생존자들을 잘 연출했으며, 레퍼런스의 창고 구조와 양개형 나무 문의 형태를 가장 일관되게 유지했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "인물들의 배치가 다소 엉성하며, 레퍼런스의 문이 아닌 Z자 형태의 단일 문으로 변형되어 장소 일관성이 떨어집니다."
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "충격받은 군상의 표정과 밀도는 좋으나, 문의 구조가 레퍼런스와 다르고 일부 남성들의 망건 형태가 다소 어색하게 왜곡되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S2sh5_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 과거의 조선인 육혈귀 생존자: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:916312>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "문 양옆에 레퍼런스 캐릭터와 완전히 동일한 얼굴, 헤어스타일, 복장을 한 여성이 두 명 나타나는 복제 오류가 발생했습니다.",
     "fix_en": "Change the face, hairstyle, and clothing details of the woman standing to the left of the door to create a distinctly different Joseon-era character, eliminating the duplicate.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "화면 왼쪽의 키 큰 남성이 맞잡고 있는 두 손의 손가락들이 서로 비정상적으로 엉켜 있고 마디가 뭉개져 있습니다.",
     "fix_en": "Redraw the clasped hands of the tall man on the left to have proper anatomical structure and a correct number of distinct human fingers.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "문 왼쪽에 서 있는 단발머리 여성의 내린 왼손(화면상 우측 손)이 손가락 형태 없이 뭉툭한 덩어리처럼 묘사되었습니다.",
     "fix_en": "Restore distinct, anatomically correct fingers to the visible left hand of the short-haired woman standing to the left of the door.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the face, hairstyle, and clothing details of the woman standing to the left of the door to create a distinctly different Joseon-era character, eliminating the duplicate.\n- Redraw the clasped hands of the tall man on the left to have proper anatomical structure and a correct number of distinct human fingers.\n- Restore distinct, anatomically correct fingers to the visible left hand of the short-haired woman standing to the left of the door.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S2sh5"
  }
 },
 "S3sh6": {
  "input_fingerprint": "d1fb9b035fd6246f",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day transitioning to night, then morning.\n\nSHOT TEXT (authoritative, Korean): 하얀 사슴뿔 모양의 긴 창이 그분(검은 연기 형체)의 가슴팍을 정면으로 꿰뚫고 등 뒤로 튀어나온 찰나.\n\nLOCATION (lock): On the bloodied ground directly in front of the reed-field storehouse. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The long spear remains driven through the black figure’s chest, with its white antler-like point projecting from his back.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 5,
   "A": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "창이 '검은 연기 형체' 중심에 위치해 관통하는 핵심 액션을 가장 근접하게 연출했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "창이 대상을 꿰뚫은 상태는 연출했으나, 지정된 연기가 아닌 검은 천 조각으로 잘못 묘사되었습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "창이 바닥에 방치되어 있고 연기와 완전히 분리되어 있어 꿰뚫는 액션을 전혀 구현하지 못했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L03B01.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "'검은 연기 형체'라는 인물이 프롬프트의 지시(모든 캐릭터는 실제 사람으로 묘사)를 위반하여 사람의 형체 없이 단순한 연기 기둥으로만 렌더링되었습니다.",
     "fix_en": "Replace the column of black smoke with a fully-formed human actor standing in the exact same spot.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "창이 인물의 가슴을 관통하여 등 뒤로 튀어나와야 한다는 묘사와 달리, 바닥에 수직으로 꽂혀 있습니다.",
     "fix_en": "Reorient the spear horizontally so it drives completely through the human actor's chest, with the white antler point protruding from their back.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "프롬프트와 레퍼런스 이미지에 명시된 '피투성이 바닥(bloodied ground)'이 묘사되지 않고 일반적인 진흙 바닥으로 표현되었습니다.",
     "fix_en": "Add prominent pools and stains of dark red blood across the muddy ground in the foreground.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the column of black smoke with a fully-formed human actor standing in the exact same spot.\n- Reorient the spear horizontally so it drives completely through the human actor's chest, with the white antler point protruding from their back.\n- Add prominent pools and stains of dark red blood across the muddy ground in the foreground.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S3sh12": {
  "input_fingerprint": "389d4ba39dd50786",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day transitioning to night, then morning.\n\nSHOT TEXT (authoritative, Korean): 목을 잡힌 채 그분(검은 연기 형체)의 얼굴 가까이 끌려가 눈을 맞추며 고개를 끄덕이듯 턱을 당긴 조선인 촌주의 정지 자세.\n\nLOCATION (lock): Beside the fallen bodies on the open ground immediately outside the reed-field storehouse. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the reed-side warehouse exterior, bloodied ground, daylight, and persistent battlefield debris. Exclude the earlier spear-through-chest instant and any figures not present in this framing.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The black figure remains impaled by the antler-tipped spear while gripping the village chief by the throat.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 그분(검은 연기 형체) (검은 연기형 얼굴, 사람형 머리 윤곽, 머리 위 두 뿔, 길고 흐릿한 신체 실루엣); 조선인 촌주 (조선인 50대 중반 남성, 상투 튼 검은 머리, 둥근 중년 얼굴, 세월이 밴 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 8,
   "A": 6,
   "C": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "촌주가 검은 형체의 얼굴 가까이 끌려가 턱을 당기고 눈을 맞추는 핵심 연출과 캐릭터의 외형을 가장 정확하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "인물 간 거리는 가깝지만 촌주가 턱을 당기는 대신 들고 있으며, 캐릭터의 외모가 레퍼런스와 다릅니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "촌주가 얼굴 가까이 끌려가지 않고 멀리 떨어져 선 상태로 묘사되어 샷의 핵심 거리감과 연출을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S3sh6_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 그분(검은 연기 형체): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1138509>"
   },
   {
    "label": "CHARACTER REFERENCE — 조선인 촌주: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1414509>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "촌주가 캐릭터 레퍼런스에 명시된 남색 한복이 아닌, 이전 샷 래퍼런스의 다른 인물이 입고 있던 낡은 회색 의상을 입고 있습니다.",
     "fix_en": "Change the village chief's clothing from the grey tunic to the dark blue patterned hanbok shown in his character reference.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "촌주의 턱이 위로 살짝 들려 있어, '턱을 당긴'이라는 프롬프트의 자세 지시와 어긋납니다.",
     "fix_en": "Tilt the village chief's head slightly down to pull his chin inward toward his neck.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the village chief's clothing from the grey tunic to the dark blue patterned hanbok shown in his character reference.\n- Tilt the village chief's head slightly down to pull his chin inward toward his neck.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S3sh6"
  }
 },
 "S3sh15": {
  "input_fingerprint": "636b8f8682c2f958",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day transitioning to night, then morning.\n\nSHOT TEXT (authoritative, Korean): 기괴한 황금빛 눈동자를 지닌 조선인 촌주가 과거의 조선인 육혈귀 생존자들을 뚫어지게 응시하는 얼굴 클로즈업.\n\nLOCATION (lock): In the clearing before the reed-field storehouse, facing the survivors emerging through its door. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the daylight look and bloodied warehouse-front environment only as the background anchor. Exclude the fallen shadow figure and the earlier impalement action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The black figure and the spear that had impaled him are gone; the slaughtered Japanese soldiers remain strewn around the warehouse.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 조선인 촌주 (그분 빙의 상태) (조선인 50대 중반 남성, 상투 튼 검은 머리, 둥근 중년 얼굴, 비인간적인 눈빛) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "C": 5,
   "B": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "얼굴 클로즈업 요건을 완벽히 충족하며, 이전 샷의 푸른 의상과 황금빛 눈동자를 정확히 유지했습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "의상 연속성은 지켰으나 프레이밍이 넓어졌고, 배경에 생존자들을 배치하여 응시한다는 시선 설정과 모순됩니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "프레이밍이 넓고, 이전 샷에서 설정된 푸른 의상 대신 베이지색 옷으로 변경된 오류가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S3sh12_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 조선인 촌주 (그분 빙의 상태): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:337038>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 샷 및 캐릭터 레퍼런스에 있던 검은색 망건과 상투가 캐릭터의 머리에 없습니다.",
     "fix_en": "Add the traditional black headband (manggeon) across his forehead and a topknot on top of his head, matching the reference.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "촌주가 창고 문에서 나오는 생존자들을 바라보고 있으므로, 그의 얼굴을 정면에서 잡은 카메라 배경에는 창고가 아닌 반대편 갈대밭이 있어야 합니다.",
     "fix_en": "Replace the warehouse in the background completely with a continuous view of the clearing and reed field, reflecting the correct camera viewpoint.",
     "needs_regeneration": false,
     "unfixable": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the traditional black headband (manggeon) across his forehead and a topknot on top of his head, matching the reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S3sh12"
  }
 },
 "S4sh1::bgfirst_bg": {
  "input_fingerprint": "3653a809420a12d4",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 가로로 긴 궤적을 그리는 도심 불빛을 배경으로, 버스 창가 자리에 앉아 피곤한 얼굴로 창밖을 바라보고 있는 김지후(강수리영)의 모습.\n\nLOCATION (lock): At a window seat inside a moving nighttime commuter bus, with city lights streaking past outside.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From seated eye level on the aisle side, the camera begins a parallel track along the window seats in a wide horizontal composition, keeping 김지후(강수리영) anchored at the window side of the frame. His seated profile and tired face remain stable in the midground while the passing city lights stretch into long horizontal trajectories beyond the bus window and other exhausted workers provide scale deeper along the aisle.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: Ji-hoo at the window seat in the middle-right of the frame, midground; passing city lights beyond the bus window in the middle-left of the frame, background.\n- KEY BACKGROUND ELEMENTS: 버스 차창 (The bus is moving) — The city lights and vehicles outside are visible through the window behind the seated figure; used as The window holds the moving city backdrop behind Ji-hoo and separates his still posture from the exterior motion; 도심 네온사인과 잡다한 불빛 (Passing rapidly outside at night); used as The lights form long horizontal trajectories that emphasize the bus's motion and urban density; 버스 좌석과 통로 (Occupied by workers who are sleeping or looking at phones) — The seat rows recede away from the camera along the aisle, with the window seats facing the direction of travel; used as The rows establish the aisle-side camera path and contain the exhausted commuters.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 가로로 긴 궤적을 그리는 도심 불빛을 배경으로, 버스 창가 자리에 앉아 피곤한 얼굴로 창밖을 바라보고 있는 김지후(강수리영)의 모습.\n\nLOCATION (lock): At a window seat inside a moving nighttime commuter bus, with city lights streaking past outside.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From seated eye level on the aisle side, the camera begins a parallel track along the window seats in a wide horizontal composition, keeping 김지후(강수리영) anchored at the window side of the frame. His seated profile and tired face remain stable in the midground while the passing city lights stretch into long horizontal trajectories beyond the bus window and other exhausted workers provide scale deeper along the aisle.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: Ji-hoo at the window seat in the middle-right of the frame, midground; passing city lights beyond the bus window in the middle-left of the frame, background.\n- KEY BACKGROUND ELEMENTS: 버스 차창 (The bus is moving) — The city lights and vehicles outside are visible through the window behind the seated figure; used as The window holds the moving city backdrop behind Ji-hoo and separates his still posture from the exterior motion; 도심 네온사인과 잡다한 불빛 (Passing rapidly outside at night); used as The lights form long horizontal trajectories that emphasize the bus's motion and urban density; 버스 좌석과 통로 (Occupied by workers who are sleeping or looking at phones) — The seat rows recede away from the camera along the aisle, with the window seats facing the direction of travel; used as The rows establish the aisle-side camera path and contain the exhausted commuters.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S4sh1__bgfirst_bg.png",
  "asset_id": "62e1f6ce-08bd-4838-a892-ba95d127a556",
  "input_asset_ids": [
   "9f0748c5-9c73-46f1-a565-16904ea033d2",
   "c69de9e8-1dfd-4264-83c3-a4f0afcc784c",
   "d74f097c-288e-4fa6-ac33-f9284505f0ba"
  ]
 },
 "S4sh1": {
  "input_fingerprint": "c57d7cc2c883ef00",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 가로로 긴 궤적을 그리는 도심 불빛을 배경으로, 버스 창가 자리에 앉아 피곤한 얼굴로 창밖을 바라보고 있는 김지후(강수리영)의 모습.\n\nLOCATION (lock): At a window seat inside a moving nighttime commuter bus, with city lights streaking past outside. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu wears her antler-bone pendant, which she has worn since her first appearance.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 가로로 긴 궤적을 그리는 도심 불빛을 배경으로, 버스 창가 자리에 앉아 피곤한 얼굴로 창밖을 바라보고 있는 김지후(강수리영)의 모습.\n\nLOCATION (lock): At a window seat inside a moving nighttime commuter bus, with city lights streaking past outside. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu wears her antler-bone pendant, which she has worn since her first appearance.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 가로로 긴 궤적을 그리는 도심 불빛을 배경으로, 버스 창가 자리에 앉아 피곤한 얼굴로 창밖을 바라보고 있는 김지후(강수리영)의 모습.\n\nLOCATION (lock): At a window seat inside a moving nighttime commuter bus, with city lights streaking past outside. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu wears her antler-bone pendant, which she has worn since her first appearance.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S4sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S4sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L11B04.png"
    },
    {
     "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_multifamily_residence_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "지시문에 없는 다른 승객들이 배경에 추가된 것은 엄격한 규칙 위반이나, 캐릭터의 외모, 지정된 의상 및 소품(펜던트), 창밖을 보는 사실적인 연출은 훌륭하게 구현되었습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "우측 버스 내부 좌석을 관통하는 물리적으로 불가능한 빛 궤적 오류가 있으며, 위치 레퍼런스의 구도를 그대로 복사하여 불필요한 휴대폰이 등장했고 캐릭터의 의상도 일치하지 않습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "지시문에 없는 다른 승객들이 배경에 추가된 것은 엄격한 규칙 위반이나, 캐릭터의 외모, 지정된 의상 및 소품(펜던트), 창밖을 보는 사실적인 연출은 훌륭하게 구현되었습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "우측 버스 내부 좌석을 관통하는 물리적으로 불가능한 빛 궤적 오류가 있으며, 위치 레퍼런스의 구도를 그대로 복사하여 불필요한 휴대폰이 등장했고 캐릭터의 의상도 일치하지 않습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "명시된 인물만 등장시키는 제한을 정확히 지켰고 펜던트와 버스 내부의 기준 요소를 잘 재현했으나, 우측 버스 실내 구조물 위로 외부의 빛 궤적이 투과되어 겹치는 물리적 렌더링 오류가 아쉽습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "주인공의 외형과 펜던트, 창밖을 보는 연출은 좋으나, 지시문에 없는 다수의 승객을 배경에 임의로 추가하여 '명시된 인물 외에는 절대 등장시키지 말 것'이라는 필수 규칙을 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "명시된 인물만 등장시키는 제한을 정확히 지켰고 펜던트와 버스 내부의 기준 요소를 잘 재현했으나, 우측 버스 실내 구조물 위로 외부의 빛 궤적이 투과되어 겹치는 물리적 렌더링 오류가 아쉽습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "주인공의 외형과 펜던트, 창밖을 보는 연출은 좋으나, 지시문에 없는 다수의 승객을 배경에 임의로 추가하여 '명시된 인물 외에는 절대 등장시키지 말 것'이라는 필수 규칙을 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 9,
     "B": 9
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 9,
   "B": 9
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "지시문에 없는 다른 승객들이 배경에 추가된 것은 엄격한 규칙 위반이나, 캐릭터의 외모, 지정된 의상 및 소품(펜던트), 창밖을 보는 사실적인 연출은 훌륭하게 구현되었습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "우측 버스 내부 좌석을 관통하는 물리적으로 불가능한 빛 궤적 오류가 있으며, 위치 레퍼런스의 구도를 그대로 복사하여 불필요한 휴대폰이 등장했고 캐릭터의 의상도 일치하지 않습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L11B04.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_multifamily_residence_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "샷 텍스트에 김지후만 명시되어 있으나 버스 뒷좌석에 다른 인물들이 임의로 추가되었습니다.",
     "fix_en": "Remove all extra passengers from the background seats, leaving only Kim Jihu in the bus.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "스케치에 따르면 고개를 창문 쪽(왼쪽)으로 완전히 돌려야 하지만, 생성된 이미지는 앞쪽을 바라보는 각도입니다.",
     "fix_en": "Turn Jihu's head to the left into a profile view so she is looking directly out the window.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all extra passengers from the background seats, leaving only Kim Jihu in the bus.\n- Turn Jihu's head to the left into a profile view so she is looking directly out the window.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S4sh1__bgfirst_bg.png",
   "bg_asset_id": "62e1f6ce-08bd-4838-a892-ba95d127a556",
   "bg_record_key": "S4sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate",
   "seed_attached": true
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S4sh4": {
  "input_fingerprint": "111c466e74cc06ad",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 핸드폰 화면 전체를 빼곡하게 채운 엄마의 절박한 메시지들과 수십 통의 부재중 통화 기록 클로즈업.\n\nLOCATION (lock): At the window seat inside the moving commuter bus, lit in part by a phone screen and passing city glow. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 5,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "신체 일부를 프레임에서 완전히 배제하여 '인물 없음' 지시를 잘 따랐으며, 절박한 메시지와 부재중 통화가 가득한 화면을 완벽히 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "화면 디테일은 우수하나 프레임에 인물의 다리(청바지)가 포함되어 '인물이나 신체 일부 배제'라는 부정 지시를 위반했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "핸드폰의 네 모서리가 모두 화면 안에 잡힌 채 허공에 떠 있는 물리적으로 불가능한 연출(Hard Violation)입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S4sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "핸드폰이 잡고 있는 손이나 지지대 없이 버스 창문 앞 허공에 비현실적으로 떠 있습니다.",
     "fix_en": "Add a hand holding the phone from the bottom, or place it resting on a physical surface.",
     "unfixable": true
    }
   ]
  },
  "fix_skipped": true,
  "fix_skip_reason": "all_issues_unfixable",
  "ref_mode": "prev만 (배경 전용·공유 계획)",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S4sh1"
  },
  "lane_policy": "share_plan_prev_bgonly"
 },
 "S4sh8": {
  "input_fingerprint": "09838e9a42417952",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 고개를 숙인 김지후(강수리영) 옆의 버스 창문 유리에 황금색 눈알을 지닌 사람 얼굴 형상의 그분(검은 연기 형체)이 흐릿하게 맺혀 있는 모습.\n\nLOCATION (lock): Beside the window of the moving nighttime bus, with the dark city sliding past the glass. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the moving night-bus interior, window seat, and streaking city lights. Exclude the earlier unobstructed window reflection and add only the required smoky face image.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 5,
   "A": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "이전 컷의 좌측 창문 구도를 올바르게 유지하며, 고개를 푹 숙인 지후와 창문에 맺힌 황금색 눈의 검은 연기 형체를 지시문 그대로 정확하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "연기 형체의 얼굴은 잘 표현되었으나 창문 위치가 이전 컷과 반대 방향이며, 지후가 고개를 숙인 자세가 다소 약하게 연출되었습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "창문 위치가 이전 컷과 반대이며, 검은 연기 형체의 '그분'이 아닌 지후 자신의 얼굴이 창문에 비쳐 지시문과 크게 어긋납니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S4sh4_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지후가 착용한 펜던트가 레퍼런스에 나타난 무광 뼈 재질이 아니라 빛나는 금속(금) 재질로 렌더링되었습니다.",
     "fix_en": "Change the pendant's material and color to matte off-white bone, removing the metallic gold reflections.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the pendant's material and color to matte off-white bone, removing the metallic gold reflections.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S4sh4"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "groupbg::냉장고_뒤편": {
  "input_fingerprint": "4dffe9a1225ac439",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "냉장고_뒤편",
    "tags": [
     "S5sh4",
     "S5sh6"
    ]
   },
   "context_sig": "504e16d846a5fd3b"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: In the narrow service space behind a convenience-store drink refrigerator, over its white tiled floor.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n퇴근 버스 내부: 빽빽한 좌석이 배치된 흔한 현대 도시의 통근 버스 내부. 네온사인과 도시 불빛이 차창 밖으로 스쳐 지나간다. (특징: 줄지어 있는 버스 좌석들; 어두운 실내 조명; 창문에 비치는 다채로운 도시 네온사인 불빛)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실내. 편의점 냉장고 - 밤\n- 각종 캔 음료들이 꽉 들어찬 냉장고 뒤편 빈 공간\n\nTIME OF DAY (lock): night.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: In the narrow service space behind a convenience-store drink refrigerator, over its white tiled floor.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n퇴근 버스 내부: 빽빽한 좌석이 배치된 흔한 현대 도시의 통근 버스 내부. 네온사인과 도시 불빛이 차창 밖으로 스쳐 지나간다. (특징: 줄지어 있는 버스 좌석들; 어두운 실내 조명; 창문에 비치는 다채로운 도시 네온사인 불빛)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실내. 편의점 냉장고 - 밤\n- 각종 캔 음료들이 꽉 들어찬 냉장고 뒤편 빈 공간\n\nTIME OF DAY (lock): night.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_냉장고_뒤편_04fee7.png",
  "asset_id": "e52822d3-ba68-4fb9-81fa-3ad023d46f09",
  "input_asset_ids": [
   "eaaad110-2657-44a4-928d-65c12f5c8b69"
  ],
  "origin_tag": "S5sh4",
  "place_text": "In the narrow service space behind a convenience-store drink refrigerator, over its white tiled floor.",
  "origin_inputs": {
   "place_text": "In the narrow service space behind a convenience-store drink refrigerator, over its white tiled floor.",
   "time_of_day_en": "night",
   "conti_asset_id": "eaaad110-2657-44a4-928d-65c12f5c8b69"
  }
 },
 "S5sh4::bgfirst_bg": {
  "input_fingerprint": "b83eec15cd0204b4",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 김지후(강수리영)가 매료된 듯 넋이 나간 표정으로 바닥에 떨어진 검붉은 핏방울을 뚫어지게 내려다보는 모습.\n\nLOCATION (lock): In the narrow service space behind a convenience-store drink refrigerator, over its white tiled floor.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the rear refrigerator space at upper-torso height, continue a restrained dolly toward 김지후(강수리영) on a three-quarter side angle, framing her vacant face prominently while tilting just enough downward to retain the dark-red blood drop near the lower edge. Use shallow focus on her transfixed expression, allowing the drop to remain a distinct but secondary point along her downward eyeline.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 흰색 타일바닥 (A single dark-red drop of blood has fallen onto it); used as Places the blood drop at the lower edge of 지후's eyeline and provides spatial contrast with her blue vest; 진열된 캔 음료 (Packed closely in the refrigerator display) — The rows present their front sides toward the refrigerator door, seen obliquely from the rear space; used as Forms layered refrigerator context behind and beside 지후 without displacing her expression as the primary focus; 폐기 음식 바구니 (Filled with food marked for disposal) — Its open top angles toward the camera, revealing discarded food and the remaining lunch boxes; used as Remains in the rear space as evidence of her interrupted meal.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 김지후(강수리영)가 매료된 듯 넋이 나간 표정으로 바닥에 떨어진 검붉은 핏방울을 뚫어지게 내려다보는 모습.\n\nLOCATION (lock): In the narrow service space behind a convenience-store drink refrigerator, over its white tiled floor.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the rear refrigerator space at upper-torso height, continue a restrained dolly toward 김지후(강수리영) on a three-quarter side angle, framing her vacant face prominently while tilting just enough downward to retain the dark-red blood drop near the lower edge. Use shallow focus on her transfixed expression, allowing the drop to remain a distinct but secondary point along her downward eyeline.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 흰색 타일바닥 (A single dark-red drop of blood has fallen onto it); used as Places the blood drop at the lower edge of 지후's eyeline and provides spatial contrast with her blue vest; 진열된 캔 음료 (Packed closely in the refrigerator display) — The rows present their front sides toward the refrigerator door, seen obliquely from the rear space; used as Forms layered refrigerator context behind and beside 지후 without displacing her expression as the primary focus; 폐기 음식 바구니 (Filled with food marked for disposal) — Its open top angles toward the camera, revealing discarded food and the remaining lunch boxes; used as Remains in the rear space as evidence of her interrupted meal.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S5sh4__bgfirst_bg.png",
  "asset_id": "2e61e631-cf19-4d7c-9a84-70c1499d98b8",
  "input_asset_ids": [
   "eaaad110-2657-44a4-928d-65c12f5c8b69",
   "e52822d3-ba68-4fb9-81fa-3ad023d46f09"
  ]
 },
 "S5sh4": {
  "input_fingerprint": "96e288eb123f6b2f",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 매료된 듯 넋이 나간 표정으로 바닥에 떨어진 검붉은 핏방울을 뚫어지게 내려다보는 모습.\n\nLOCATION (lock): In the narrow service space behind a convenience-store drink refrigerator, over its white tiled floor. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 매료된 듯 넋이 나간 표정으로 바닥에 떨어진 검붉은 핏방울을 뚫어지게 내려다보는 모습.\n\nLOCATION (lock): In the narrow service space behind a convenience-store drink refrigerator, over its white tiled floor. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 매료된 듯 넋이 나간 표정으로 바닥에 떨어진 검붉은 핏방울을 뚫어지게 내려다보는 모습.\n\nLOCATION (lock): In the narrow service space behind a convenience-store drink refrigerator, over its white tiled floor. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S5sh4__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S5sh4.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_냉장고_뒤편_04fee7.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지후가 바닥의 핏방울을 내려다보는 넋 나간 표정과 지정된 장소, 펜던트 소품을 지시문대로 잘 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "화면 좌측 하단에 바닥과 핏방울이 별도의 패널/콜라주 형태로 삽입되어 하드 비올레이션 규정을 위반했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지후가 바닥의 핏방울을 내려다보는 넋 나간 표정과 지정된 장소, 펜던트 소품을 지시문대로 잘 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "화면 좌측 하단에 바닥과 핏방울이 별도의 패널/콜라주 형태로 삽입되어 하드 비올레이션 규정을 위반했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 장소에서 바닥의 핏자국을 내려다보는 인물의 표정과 자세, 펜던트 착용 상태를 정확히 묘사했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "좌측 하단에 바닥 화면이 부자연스럽게 합성된 콜라주 형태이므로 치명적인 규정 위반입니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 장소에서 바닥의 핏자국을 내려다보는 인물의 표정과 자세, 펜던트 착용 상태를 정확히 묘사했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "좌측 하단에 바닥 화면이 부자연스럽게 합성된 콜라주 형태이므로 치명적인 규정 위반입니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 6
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지후가 바닥의 핏방울을 내려다보는 넋 나간 표정과 지정된 장소, 펜던트 소품을 지시문대로 잘 구현했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "화면 좌측 하단에 바닥과 핏방울이 별도의 패널/콜라주 형태로 삽입되어 하드 비올레이션 규정을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_냉장고_뒤편_04fee7.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지후의 팔뚝이 맨살로 드러나 있습니다. 캐릭터 레퍼런스와 같이 스웨터 소매가 손목까지 내려와 팔을 덮어야 합니다.",
     "fix_en": "Extend the dark green sweater sleeves all the way down to the wrists, covering the exposed bare forearms to match the character reference outfit.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Extend the dark green sweater sleeves all the way down to the wrists, covering the exposed bare forearms to match the character reference outfit.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S5sh4__bgfirst_bg.png",
   "bg_asset_id": "2e61e631-cf19-4d7c-9a84-70c1499d98b8",
   "bg_record_key": "S5sh4::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "냉장고_뒤편",
   "groupbg_asset_id": "e52822d3-ba68-4fb9-81fa-3ad023d46f09"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S5sh6": {
  "input_fingerprint": "09c309cf0f167816",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 음료수가 진열된 냉장고 유리문 바깥쪽에서 정인우가 김지후(강수리영)를 향해 환한 미소를 짓고 있는 얼굴.\n\nLOCATION (lock): At the glass drink-refrigerator doors separating the rear service space from the convenience-store sales floor. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the packed drink refrigerator, glass door, white-tiled floor, and cold store lighting. Exclude the blood-drop fixation moment and frame the man only outside the glass.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The dark-red drop of Jihu’s blood remains on the white tile floor, and Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "이전 샷의 공간 방향을 정확히 유지하면서 유리문 바깥에서 환하게 웃는 남성의 모습을 샷의 의도와 물리적 반사에 맞게 잘 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "지후의 시선과 위치가 레퍼런스와 반대로 뒤집혔으며, 물리적으로 불가능하게 허공에 떠 있는 펜던트 반사 이미지가 나타나 감점되었습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "인물의 공간적 방향이 반전되었고, 배경에 프롬프트에 없는 기괴한 얼굴이 추가되었으며 바닥 핏방울이 콜라주처럼 단순 합성되어 하드 위반입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S5sh4_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1225321>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "남자의 얼굴과 상체가 반투명하게 렌더링되어, 뒤쪽에 있는 음료 캔과 선반이 그의 몸을 투과하여 보입니다.",
     "fix_en": "Make the man's face and upper body completely solid and opaque, eliminating all transparency so the background shelves and drinks do not show through him.",
     "unfixable": false
    },
    {
     "issue_ko": "카메라가 쿨러 내부에서 바깥을 향해 있어야 하므로 진열된 음료의 뒷면이 보여야 하지만, 음료의 정면과 내부 바닥이 왜곡되어 겹쳐 보이는 등 공간 기하학이 완전히 무너졌습니다.",
     "fix_en": "The spatial geometry is fundamentally broken, showing the front of the drinks and blending multiple perspectives; this cannot be repaired without changing the camera viewpoint and rebuilding the scene.",
     "unfixable": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Make the man's face and upper body completely solid and opaque, eliminating all transparency so the background shelves and drinks do not show through him.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S5sh4"
  }
 },
 "S6sh3::bgfirst_bg": {
  "input_fingerprint": "f38387bb09c60a48",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 한 손에 캔커피를 쥔 채 까만 밤하늘을 올려다보며 입술을 동그랗게 모으고 있는 김지후(강수리영)의 측면.\n\nLOCATION (lock): On a bench directly outside a convenience store at night, beneath the open sky.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: End the dolly-in beside the bench at seated eye level, isolating 김지후(강수리영)'s right-facing side profile while leaving a small field of open night sky above her lifted eyeline. Her rounded lips and raised chin carry the focal emphasis, with the opened canned coffee retained low in frame as a compact scale reference.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 벤치 (Occupied by 지후 and 정인우) — Its seating line runs laterally beneath 지후 along the conversational axis; used as Supports 지후's seated posture along the lower boundary without competing with her profile; 캔커피 (Opened and being held by 지후) — The opened top angles slightly upward in 지후's hand while the cylindrical side remains visible; used as Held low in the composition to preserve the preceding action and identify the casual pause before the warning; 밤하늘 (Dark at night); used as Provides clear negative space in the direction of 지후's upward gaze.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 한 손에 캔커피를 쥔 채 까만 밤하늘을 올려다보며 입술을 동그랗게 모으고 있는 김지후(강수리영)의 측면.\n\nLOCATION (lock): On a bench directly outside a convenience store at night, beneath the open sky.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: End the dolly-in beside the bench at seated eye level, isolating 김지후(강수리영)'s right-facing side profile while leaving a small field of open night sky above her lifted eyeline. Her rounded lips and raised chin carry the focal emphasis, with the opened canned coffee retained low in frame as a compact scale reference.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 벤치 (Occupied by 지후 and 정인우) — Its seating line runs laterally beneath 지후 along the conversational axis; used as Supports 지후's seated posture along the lower boundary without competing with her profile; 캔커피 (Opened and being held by 지후) — The opened top angles slightly upward in 지후's hand while the cylindrical side remains visible; used as Held low in the composition to preserve the preceding action and identify the casual pause before the warning; 밤하늘 (Dark at night); used as Provides clear negative space in the direction of 지후's upward gaze.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S6sh3__bgfirst_bg.png",
  "asset_id": "e7a1eb43-c872-4d27-800a-58977bc70ba6",
  "input_asset_ids": [
   "4fbe6161-68af-4ba8-8518-4d173727419f",
   "8a7d6b9f-916c-4871-b7d2-3bf4872c547f"
  ]
 },
 "S6sh3": {
  "input_fingerprint": "9edb82377b1d1a99",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 한 손에 캔커피를 쥔 채 까만 밤하늘을 올려다보며 입술을 동그랗게 모으고 있는 김지후(강수리영)의 측면.\n\nLOCATION (lock): On a bench directly outside a convenience store at night, beneath the open sky. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and holds the canned coffee Inwoo gave her.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 한 손에 캔커피를 쥔 채 까만 밤하늘을 올려다보며 입술을 동그랗게 모으고 있는 김지후(강수리영)의 측면.\n\nLOCATION (lock): On a bench directly outside a convenience store at night, beneath the open sky. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and holds the canned coffee Inwoo gave her.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 한 손에 캔커피를 쥔 채 까만 밤하늘을 올려다보며 입술을 동그랗게 모으고 있는 김지후(강수리영)의 측면.\n\nLOCATION (lock): On a bench directly outside a convenience store at night, beneath the open sky. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and holds the canned coffee Inwoo gave her.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S6sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S6sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L06B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시된 표정과 소품을 잘 구현했으나, 벤치 위치가 원본과 달리 창문 앞으로 변경됨."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "전경에 존재하지 않는 나무 벤치를 추가로 생성하여 하드 위반(없는 사물 창조)."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시된 표정과 소품을 잘 구현했으나, 벤치 위치가 원본과 달리 창문 앞으로 변경됨."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "전경에 존재하지 않는 나무 벤치를 추가로 생성하여 하드 위반(없는 사물 창조)."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "장소의 실제 공간 구조(코너와 유리창의 위치)를 기준과 완벽하게 일치시켰으며, 요구된 포즈와 소품 배치도 정확합니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "인물의 포즈와 소품은 잘 표현되었으나, 기준 이미지의 건축 구조(벽면과 편의점 유리창의 배치)를 왜곡하여 장소 일치도가 떨어집니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "장소의 실제 공간 구조(코너와 유리창의 위치)를 기준과 완벽하게 일치시켰으며, 요구된 포즈와 소품 배치도 정확합니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "인물의 포즈와 소품은 잘 표현되었으나, 기준 이미지의 건축 구조(벽면과 편의점 유리창의 배치)를 왜곡하여 장소 일치도가 떨어집니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 13
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 11,
   "B": 13
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지시된 표정과 소품을 잘 구현했으나, 벤치 위치가 원본과 달리 창문 앞으로 변경됨."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "전경에 존재하지 않는 나무 벤치를 추가로 생성하여 하드 위반(없는 사물 창조)."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L06B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "벤치가 레퍼런스의 막힌 벽면이 아닌 상점 유리창 앞에 잘못 배치되어 장소의 구조와 어긋납니다.",
     "fix_en": "The bench is incorrectly placed in front of the glass storefront instead of the solid wall recess shown in the reference.",
     "needs_regeneration": true,
     "unfixable": true
    },
    {
     "issue_ko": "목걸이 펜던트가 레퍼런스의 매끄러운 뼈 재질이 아닌 복잡하게 조각된 금속 장식 모양으로 묘사되었습니다.",
     "fix_en": "Replace the intricate pendant with the simple, smooth bone arrowhead pendant from the character reference.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "캔커피, 배경의 파란색 표지판 및 상점 유리창 포스터에 의미 없는 텍스트가 생성되어 텍스트 금지 지시를 위반했습니다.",
     "fix_en": "Remove all garbled text from the coffee can, the blue sign, and the posters in the window.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the intricate pendant with the simple, smooth bone arrowhead pendant from the character reference.\n- Remove all garbled text from the coffee can, the blue sign, and the posters in the window.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S6sh3__bgfirst_bg.png",
   "bg_asset_id": "e7a1eb43-c872-4d27-800a-58977bc70ba6",
   "bg_record_key": "S6sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S6sh4": {
  "input_fingerprint": "798d86d93bce874c",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 갑작스러운 휘파람 소리에 놀란 정인우가 정색하며 김지후(강수리영)를 똑바로 쳐다보는 옆얼굴.\n\nLOCATION (lock): On the nighttime bench immediately outside the convenience store. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the nighttime storefront bench, exterior lighting, and seated arrangement. Exclude the woman's upward-looking whistle pose and emphasize the man's startled reaction.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and keeps the opened canned coffee in hand.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 4,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "정인우가 김지후를 똑바로 쳐다보는 옆얼굴 프레이밍과 시선, 그리고 지후의 소품(캔커피, 펜던트) 유지 조건을 모두 완벽하게 충족했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "지후의 소품 상태는 유지되었으나, 인우가 지후를 쳐다보지 않고 엉뚱한 곳을 응시하여 샷 텍스트의 핵심 시선 연출을 위반했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "인우가 지후를 향해 시선을 주지 않으며 프레이밍도 지나치게 넓어져 '쳐다보는 옆얼굴'이라는 지시를 제대로 구현하지 못했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S6sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:755998>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "커피 캔을 쥐고 있는 여자의 오른손 손가락들이 비정상적으로 융합되거나 갯수가 왜곡되어 해부학적으로 어색합니다.",
     "fix_en": "Redraw the woman's right hand holding the coffee can to have five anatomically correct fingers grasping the cylinder naturally.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the woman's right hand holding the coffee can to have five anatomically correct fingers grasping the cylinder naturally.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S6sh3"
  }
 },
 "S6sh7": {
  "input_fingerprint": "266d6a2d6ab23767",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 미간을 찌푸린 불쾌한 표정으로 정인우를 매섭게 쏘아보는 김지후(강수리영)의 얼굴.\n\nLOCATION (lock): At the bench area in front of the convenience store, exposed to the night street. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the nighttime storefront, bench, and ambient street lighting. Exclude the man's earlier startled pose and any coffee action not visible in the close framing.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and retains the canned coffee.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지후의 단독 얼굴 샷을 정확히 구현했으며, 필수 소품(펜던트, 캔커피)과 오프스크린을 향한 시선 방향을 훌륭하게 유지했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "지후의 단독 샷이나 필수 소품인 펜던트가 누락되었고, 시선이 정면을 향해 이전 샷과의 공간적 연속성이 어긋납니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "허용된 인물 목록에 없는 인물(정인우)의 뒷모습을 포함하여 인물 제한 규정을 명백히 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S6sh4_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "커피 캔의 디자인이 이전 장면과 일치하지 않습니다(이전 장면은 윗부분이 짙은 갈색, 아랫부분이 베이지색인 투톤입니다). 또한 캔이 열려 있고 금지된 텍스트가 표기되어 있습니다.",
     "fix_en": "Recolor the bottom half of the coffee can to light beige, remove the text, and make the top flat and closed to match the previous still.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "목걸이 펜던트의 형태가 레퍼런스 이미지의 날카로운 화살촉 모양과 다르게 울퉁불퉁한 형태로 변형되어 있습니다.",
     "fix_en": "Reshape the pendant into the sharp, carved triangular arrowhead seen in the character reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Recolor the bottom half of the coffee can to light beige, remove the text, and make the top flat and closed to match the previous still.\n- Reshape the pendant into the sharp, carved triangular arrowhead seen in the character reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S6sh4"
  }
 },
 "S7sh1::bgfirst_bg": {
  "input_fingerprint": "486e27ea394ffed8",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 밝은 대낮의 경찰서 주차장에서 강민숙(홍은미)이 근심이 가득한 얼굴로 정인우의 눈을 똑바로 마주 보고 선 모습.\n\nLOCATION (lock): In an open police-station parking lot during daylight, where the two stand face to face.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level and a balanced wide distance, hold a static perpendicular two-person view with 강민숙(홍은미) on the left and 정인우 on the right, their opposing profiles defining the confrontation across the center. 민숙 leans subtly into the exchange while 인우's weight shifts away, allowing their unequal urgency to read through posture before the later push toward her face.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 경찰서 주차장 (Seen during daytime); used as Establishes the open public setting and leaves clear negative space around the private confrontation; 경찰서 외부 (Visible as the setting boundary) — The exterior faces toward the parking area behind the two figures; used as Locates the conversation outside the police station without distracting from the balanced figures.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 밝은 대낮의 경찰서 주차장에서 강민숙(홍은미)이 근심이 가득한 얼굴로 정인우의 눈을 똑바로 마주 보고 선 모습.\n\nLOCATION (lock): In an open police-station parking lot during daylight, where the two stand face to face.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level and a balanced wide distance, hold a static perpendicular two-person view with 강민숙(홍은미) on the left and 정인우 on the right, their opposing profiles defining the confrontation across the center. 민숙 leans subtly into the exchange while 인우's weight shifts away, allowing their unequal urgency to read through posture before the later push toward her face.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 경찰서 주차장 (Seen during daytime); used as Establishes the open public setting and leaves clear negative space around the private confrontation; 경찰서 외부 (Visible as the setting boundary) — The exterior faces toward the parking area behind the two figures; used as Locates the conversation outside the police station without distracting from the balanced figures.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh1__bgfirst_bg.png",
  "asset_id": "e8ca62b1-5fb0-40ad-b49d-77abc1e520bf",
  "input_asset_ids": [
   "dfb060e3-ce22-4d7b-801c-f5e3cedd0f97",
   "23616602-107a-42b0-b62e-3e68ee80b371"
  ]
 },
 "S7sh1": {
  "input_fingerprint": "07c4c000587171e4",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 대낮의 경찰서 주차장에서 강민숙(홍은미)이 근심이 가득한 얼굴로 정인우의 눈을 똑바로 마주 보고 선 모습.\n\nLOCATION (lock): In an open police-station parking lot during daylight, where the two stand face to face. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 대낮의 경찰서 주차장에서 강민숙(홍은미)이 근심이 가득한 얼굴로 정인우의 눈을 똑바로 마주 보고 선 모습.\n\nLOCATION (lock): In an open police-station parking lot during daylight, where the two stand face to face. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 대낮의 경찰서 주차장에서 강민숙(홍은미)이 근심이 가득한 얼굴로 정인우의 눈을 똑바로 마주 보고 선 모습.\n\nLOCATION (lock): In an open police-station parking lot during daylight, where the two stand face to face. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S7sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:755998>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L08B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:755998>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "배경의 표지판과 차량 배치를 포함해 지정된 장소를 매우 정확하게 재현했으며, 두 인물이 마주 보고 선 구도와 표정 지시도 충실히 따랐습니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "인물의 표정과 시선 처리는 훌륭하나, 레퍼런스 이미지에 명확히 존재하는 벽면의 파란색 주차 표지판이 누락되어 장소 구현의 정확도가 떨어집니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "배경의 표지판과 차량 배치를 포함해 지정된 장소를 매우 정확하게 재현했으며, 두 인물이 마주 보고 선 구도와 표정 지시도 충실히 따랐습니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "인물의 표정과 시선 처리는 훌륭하나, 레퍼런스 이미지에 명확히 존재하는 벽면의 파란색 주차 표지판이 누락되어 장소 구현의 정확도가 떨어집니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "요구된 근심 가득한 표정과 시선을 적절한 구도로 잘 살렸으며, 위치 레퍼런스의 표지판 디테일까지 정확하게 재현했습니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "넓은 샷으로 인해 표정 묘사가 덜 부각되며, 배경의 표지판 형태가 변형되어 위치 일치도가 상대적으로 떨어집니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "요구된 근심 가득한 표정과 시선을 적절한 구도로 잘 살렸으며, 위치 레퍼런스의 표지판 디테일까지 정확하게 재현했습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "넓은 샷으로 인해 표정 묘사가 덜 부각되며, 배경의 표지판 형태가 변형되어 위치 일치도가 상대적으로 떨어집니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 13
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 13
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "배경의 표지판과 차량 배치를 포함해 지정된 장소를 매우 정확하게 재현했으며, 두 인물이 마주 보고 선 구도와 표정 지시도 충실히 따랐습니다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "인물의 표정과 시선 처리는 훌륭하나, 레퍼런스 이미지에 명확히 존재하는 벽면의 파란색 주차 표지판이 누락되어 장소 구현의 정확도가 떨어집니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L08B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:755998>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "강민숙이 정인우의 눈을 똑바로 마주보지 않고 가슴이나 목 부근을 바라보고 있으며, 정인우의 시선 역시 어긋나 있습니다. (프롬프트 텍스트 및 스케치 화살표 방향 위반)",
     "fix_en": "Adjust the eyes and head angles of both characters so they are looking directly into each other's eyes.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "배경이 원본과 동일하게 유지되지 않았습니다. 벽면의 원형 표지판(숫자 33)과 파란색 안내 표지판들(벽면 및 스탠드)의 텍스트가 알 수 없는 문자로 변형되었습니다.",
     "fix_en": "Restore the number '33' on the circular wall sign and the original Korean text on both blue parking signs to match the background reference exactly.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Adjust the eyes and head angles of both characters so they are looking directly into each other's eyes.\n- Restore the number '33' on the circular wall sign and the original Korean text on both blue parking signs to match the background reference exactly.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh1__bgfirst_bg.png",
   "bg_asset_id": "e8ca62b1-5fb0-40ad-b49d-77abc1e520bf",
   "bg_record_key": "S7sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S7sh6": {
  "input_fingerprint": "329d127ba6c56d3f",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 강민숙(홍은미)의 손이 두툼한 돈봉투를 당황한 정인우의 펼쳐진 손바닥 위에 강하게 억누르고 있어 정인우의 손바닥이 살짝 눌린 클로즈업 찰나.\n\nLOCATION (lock): In the daylight parking area outside the police station. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the bright daytime police-station parking-lot look and the two speakers' continuing clothing. Exclude their earlier face-to-face pose and focus only on the envelope being pressed into the hand.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The thick envelope of money is transferred into Inwoo’s hand and remains in his possession.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 5,
   "A": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "요구된 클로즈업 프레이밍에 정확히 부합하며, 봉투를 강하게 억누르는 손의 동작과 위치가 가장 자연스럽게 연출되었습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "지시된 타이트한 클로즈업보다 다소 넓게 잡혔으며, 손바닥으로 누르기보다 손가락으로 쥐고 있는 형태입니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "아래쪽 손의 손가락 형태가 해부학적으로 어색하며, 위쪽 손이 봉투를 누르지 않고 주먹을 쥔 모양이라 지시문과 어긋납니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:755998>"
   },
   {
    "label": "PROP REFERENCE — 두툼한 돈봉투: the exact object appearing in this shot; match its look, material and wear exactly.",
    "path": "<bytes:1264547>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "돈봉투에 레퍼런스 이미지의 상단 테이프 밀봉과 측면의 '10000' 지폐 인쇄 디테일이 없습니다.",
     "fix_en": "Add the tape seal across the top flap and the faded '10000' bill graphic to the side of the brown paper envelope to match the reference prop.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the tape seal across the top flap and the faded '10000' bill graphic to the side of the brown paper envelope to match the reference prop.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S7sh1"
  }
 },
 "S7sh8::bgfirst_bg": {
  "input_fingerprint": "f173e913a1459499",
  "prompt": "Turn the attached storyboard sketch into a photorealistic film still of\nits LOCATION ONLY, keeping the figures exactly as they are.\n\nKEEP EXACTLY: the camera framing, the horizon, and where every element\nsits in the frame. Each mannequin figure stays a plain grey featureless\nmannequin standing in the very same spot, at the same size, in the same\npose, turned the same way — do not turn them into people, do not move,\nrotate, mirror or re-pose them, do not add or remove figures.\n\nBUILD PHOTOREALISTICALLY: everything that is not a figure — ground,\nsurfacing, structures, vegetation, sky, water, distance. The bare lines\nof the sketch are a layout guide; replace them with the real materials,\ndepth and lighting of the place described below.\n\nSHOT TEXT this background must serve (Korean): 어깨가 축 처진 채 주차장을 걷는 mid-stride 자세로 고정된 강민숙(홍은미)의 뒷모습과, 그 자리에 우두커니 서서 그를 바라보는 정인우.\n\nLOCATION (lock): Across the open police-station parking lot, along the woman's path away from the parked vehicles.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Complete the dolly-out at eye level behind and slightly beside 정인우, placing his arrested over-shoulder figure in the right foreground while 강민숙(홍은미) recedes mid-stride toward the left background. Hold wide once the depth between them becomes unmistakable, using their separation along the parking-lot path as the final emotional image of the memory.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 정인우 in the middle-right of the frame, foreground, looks toward 강민숙 walking away; 강민숙(홍은미) in the middle-left of the frame, background, moves toward far side of the parking lot.\n- KEY BACKGROUND ELEMENTS: 경찰서 주차장 (Seen in daylight as 민숙 walks away); used as Provides the visible departure path that separates 민숙 in depth from the stationary 인우; 경찰서 외부 (Visible during the recalled daytime scene) — The exterior is seen beyond and to the side of the departure path; used as Forms the distant spatial boundary toward which 민숙 departs.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nTHINGS AT THIS PLACE:\nThe site is organized around a police station parking area that forms the principal camera-accessible outdoor zone. A vehicular approach directly meets one edge of the parking area and connects it to the off-site road network. The source does not establish a fixed compass orientation, a visible police station facade, marked pedestrian paths, or other permanent structures, so those details remain unspecified.\n- parking lot: police station parking area\n- approach: vehicular approach to the police station parking area\n\nWORLD FACTS (creator-confirmed — always true):\n- Region (real-world reference): 대한민국 서울권과 서남해안 일대, 대한민국 영토 내 서해의 가상 외딴섬 공동체\n- Era: 주 배경은 스마트폰과 현대 경찰·교통 체계가 존재하는 2020년대 중반의 현대이며, 1670년 조선시대와 20세기 전반 일제강점기 전후의 역사적 재현 구간이 명확히 구분되어 병존한다.\n- 현대 구간은 2020년대 대한민국의 생활상으로, 1670년대 구간은 조선 후기의 복식과 생활 도구로, 20세기 전반 구간은 당시 조선인 복식과 일본군 장비로 표현한다. 시대 전환이 없는 프레임에는 서로 다른 시대의 의상·무기·교통수단·전자기기를 혼합하지 않는다.\n- 현대 육지에서는 스마트폰, 메신저 앱, 지도 앱, CCTV, 바코드 출근 시스템, 현대 경찰 장비, 승용차·구형 SUV·동력 어선이 자연스럽게 존재한다. 섬에서는 휴대전화 신호와 인터넷 연결이 불안정하거나 차단되며, 제한적인 전기 조명과 오래된 생활 설비를 우선하되 현대 물품 자체가 전혀 없는 원시사회로 오해하지 않는다. 역사 구간에는 디지털 기기와 현대 차량을 배제한다.\n\nTHINGS THAT LIVE AT THIS PLACE (what they are, not where they go): the\nlist above names what permanently belongs to this site. Where each one\nsits in this frame is decided by the attached sketch alone — do not add\nor reposition anything on the strength of the list, build only what the\nsketch already has lines for, and do not invent facilities that are not\nlisted.\n\nNo text, no captions, no diagram symbols, no markers anywhere in the\nimage.",
  "effective_prompt": "Turn the attached storyboard sketch into a photorealistic film still of\nits LOCATION ONLY, keeping the figures exactly as they are.\n\nKEEP EXACTLY: the camera framing, the horizon, and where every element\nsits in the frame. Each mannequin figure stays a plain grey featureless\nmannequin standing in the very same spot, at the same size, in the same\npose, turned the same way — do not turn them into people, do not move,\nrotate, mirror or re-pose them, do not add or remove figures.\n\nBUILD PHOTOREALISTICALLY: everything that is not a figure — ground,\nsurfacing, structures, vegetation, sky, water, distance. The bare lines\nof the sketch are a layout guide; replace them with the real materials,\ndepth and lighting of the place described below.\n\nSHOT TEXT this background must serve (Korean): 어깨가 축 처진 채 주차장을 걷는 mid-stride 자세로 고정된 강민숙(홍은미)의 뒷모습과, 그 자리에 우두커니 서서 그를 바라보는 정인우.\n\nLOCATION (lock): Across the open police-station parking lot, along the woman's path away from the parked vehicles.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Complete the dolly-out at eye level behind and slightly beside 정인우, placing his arrested over-shoulder figure in the right foreground while 강민숙(홍은미) recedes mid-stride toward the left background. Hold wide once the depth between them becomes unmistakable, using their separation along the parking-lot path as the final emotional image of the memory.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 정인우 in the middle-right of the frame, foreground, looks toward 강민숙 walking away; 강민숙(홍은미) in the middle-left of the frame, background, moves toward far side of the parking lot.\n- KEY BACKGROUND ELEMENTS: 경찰서 주차장 (Seen in daylight as 민숙 walks away); used as Provides the visible departure path that separates 민숙 in depth from the stationary 인우; 경찰서 외부 (Visible during the recalled daytime scene) — The exterior is seen beyond and to the side of the departure path; used as Forms the distant spatial boundary toward which 민숙 departs.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nTHINGS AT THIS PLACE:\nThe site is organized around a police station parking area that forms the principal camera-accessible outdoor zone. A vehicular approach directly meets one edge of the parking area and connects it to the off-site road network. The source does not establish a fixed compass orientation, a visible police station facade, marked pedestrian paths, or other permanent structures, so those details remain unspecified.\n- parking lot: police station parking area\n- approach: vehicular approach to the police station parking area\n\nWORLD FACTS (creator-confirmed — always true):\n- Region (real-world reference): 대한민국 서울권과 서남해안 일대, 대한민국 영토 내 서해의 가상 외딴섬 공동체\n- Era: 주 배경은 스마트폰과 현대 경찰·교통 체계가 존재하는 2020년대 중반의 현대이며, 1670년 조선시대와 20세기 전반 일제강점기 전후의 역사적 재현 구간이 명확히 구분되어 병존한다.\n- 현대 구간은 2020년대 대한민국의 생활상으로, 1670년대 구간은 조선 후기의 복식과 생활 도구로, 20세기 전반 구간은 당시 조선인 복식과 일본군 장비로 표현한다. 시대 전환이 없는 프레임에는 서로 다른 시대의 의상·무기·교통수단·전자기기를 혼합하지 않는다.\n- 현대 육지에서는 스마트폰, 메신저 앱, 지도 앱, CCTV, 바코드 출근 시스템, 현대 경찰 장비, 승용차·구형 SUV·동력 어선이 자연스럽게 존재한다. 섬에서는 휴대전화 신호와 인터넷 연결이 불안정하거나 차단되며, 제한적인 전기 조명과 오래된 생활 설비를 우선하되 현대 물품 자체가 전혀 없는 원시사회로 오해하지 않는다. 역사 구간에는 디지털 기기와 현대 차량을 배제한다.\n\nTHINGS THAT LIVE AT THIS PLACE (what they are, not where they go): the\nlist above names what permanently belongs to this site. Where each one\nsits in this frame is decided by the attached sketch alone — do not add\nor reposition anything on the strength of the list, build only what the\nsketch already has lines for, and do not invent facilities that are not\nlisted.\n\nNo text, no captions, no diagram symbols, no markers anywhere in the\nimage.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh8__bgfirst_bg.png",
  "asset_id": "aa5ae64e-3e09-4af0-9296-87fac507ce82",
  "input_asset_ids": [
   "7ac61eff-4098-41f7-8ed3-1eaa33464942"
  ]
 },
 "S7sh8": {
  "input_fingerprint": "a66d78c57fa027b1",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어깨가 축 처진 채 주차장을 걷는 mid-stride 자세로 고정된 강민숙(홍은미)의 뒷모습과, 그 자리에 우두커니 서서 그를 바라보는 정인우.\n\nLOCATION (lock): Across the open police-station parking lot, along the woman's path away from the parked vehicles. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the daytime parking lot, building-side surroundings, and consistent light. Exclude the money handoff and place the woman walking away while the man remains behind.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo still holds the thick envelope of money as Minsuk walks away.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Replace every grey or outlined mannequin figure in the FIRST attached\nimage with the real people the CHARACTER REFERENCE images show, and\noutput a photorealistic film still. Each mannequin becomes the character\nwhose POSE CANON and the shot text match that figure's pose and\nposition; never repeat one person across two mannequins. If no CHARACTER\nREFERENCE image is attached for a figure, still replace it with a\nplausible real person consistent with the text. Every mannequin becomes\na person — erasing a figure, or leaving one out, is not a replacement,\nand the number of figures never changes.\n\nKEEP EXACTLY: the background of the first image, the camera framing,\nand each mannequin's position, size, pose and the direction it is\nturned. A person must be exactly where their mannequin was, at the same\nscale, in the same pose, facing the same way — never mirrored, never\nre-staged, never straightened up, stood upright or otherwise re-posed.\nNo mannequin, grey figure or sketch line may remain anywhere in the\noutput.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어깨가 축 처진 채 주차장을 걷는 mid-stride 자세로 고정된 강민숙(홍은미)의 뒷모습과, 그 자리에 우두커니 서서 그를 바라보는 정인우.\n\nLOCATION (lock): Across the open police-station parking lot, along the woman's path away from the parked vehicles. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo still holds the thick envelope of money as Minsuk walks away.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Replace every grey or outlined mannequin figure in the FIRST attached\nimage with the real people the CHARACTER REFERENCE images show, and\noutput a photorealistic film still. Each mannequin becomes the character\nwhose POSE CANON and the shot text match that figure's pose and\nposition; never repeat one person across two mannequins. If no CHARACTER\nREFERENCE image is attached for a figure, still replace it with a\nplausible real person consistent with the text. Every mannequin becomes\na person — erasing a figure, or leaving one out, is not a replacement,\nand the number of figures never changes.\n\nKEEP EXACTLY: the background of the first image, the camera framing,\nand each mannequin's position, size, pose and the direction it is\nturned. A person must be exactly where their mannequin was, at the same\nscale, in the same pose, facing the same way — never mirrored, never\nre-staged, never straightened up, stood upright or otherwise re-posed.\nNo mannequin, grey figure or sketch line may remain anywhere in the\noutput.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어깨가 축 처진 채 주차장을 걷는 mid-stride 자세로 고정된 강민숙(홍은미)의 뒷모습과, 그 자리에 우두커니 서서 그를 바라보는 정인우.\n\nLOCATION (lock): Across the open police-station parking lot, along the woman's path away from the parked vehicles. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo still holds the thick envelope of money as Minsuk walks away.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "C": "Replace every grey or outlined mannequin figure in the FIRST attached\nimage with the real people the CHARACTER REFERENCE images show, and\noutput a photorealistic film still. Each mannequin becomes the character\nwhose POSE CANON and the shot text match that figure's pose and\nposition; never repeat one person across two mannequins. If no CHARACTER\nREFERENCE image is attached for a figure, still replace it with a\nplausible real person consistent with the text. Every mannequin becomes\na person — erasing a figure, or leaving one out, is not a replacement,\nand the number of figures never changes.\n\nKEEP EXACTLY: the background of the first image, the camera framing,\nand each mannequin's position, size, pose and the direction it is\nturned. A person must be exactly where their mannequin was, at the same\nscale, in the same pose, facing the same way — never mirrored, never\nre-staged, never straightened up, stood upright or otherwise re-posed.\nNo mannequin, grey figure or sketch line may remain anywhere in the\noutput.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어깨가 축 처진 채 주차장을 걷는 mid-stride 자세로 고정된 강민숙(홍은미)의 뒷모습과, 그 자리에 우두커니 서서 그를 바라보는 정인우.\n\nLOCATION (lock): Across the open police-station parking lot, along the woman's path away from the parked vehicles. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo still holds the thick envelope of money as Minsuk walks away.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh8__bgfirst_bg.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:755998>"
    }
   ],
   "B": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh8__bgfirst_bg.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:755998>"
    }
   ],
   "C": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh8__bgfirst_bg.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:755998>"
    }
   ]
  },
  "totals": {
   "C": 7,
   "B": 5,
   "A": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "레퍼런스의 인물 배치와 뒷모습 자세를 정확히 유지했으며, 지시된 두툼한 돈 봉투 소품도 가장 명확하게 재현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "구도와 자세는 레퍼런스를 잘 따랐으나, 손에 든 물건이 두툼한 돈 봉투의 형태를 띠고 있지 않습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "레퍼런스에서 확립된 남성의 뒷모습 자세를 따르지 않고 측면으로 서 있어 샷의 일관성을 훼손했습니다."
   }
  ],
  "refs": [
   {
    "label": "SHOT BACKGROUND",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh8__bgfirst_bg.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:755998>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "오른쪽 남성의 고개가 왼쪽으로 돌아가 있습니다. 마네킹처럼 고개가 정면(배경 방향)을 똑바로 향하도록 방향을 수정해야 합니다.",
     "fix_en": "Turn the man's head to face straight forward toward the background, perfectly matching the mannequin's original head direction."
    },
    {
     "issue_ko": "오른쪽 남성의 왼팔이 앞으로 굽혀져 있습니다. 양팔을 옆으로 곧게 내린 마네킹의 자세를 정확히 유지한 채 손에 봉투를 쥐도록 수정해야 합니다.",
     "fix_en": "Straighten the man's left arm to hang straight down at his side, holding the envelope next to his leg without bending the elbow."
    },
    {
     "issue_ko": "봉투를 쥔 남성의 왼손과 지폐가 엉켜 해부학적으로 어색하며, 특히 엄지손가락이 바깥쪽(왼쪽)을 향하는 기형적인 형태입니다.",
     "fix_en": "Reconstruct the left hand and envelope to have correct human anatomy, properly positioning the thumb on the inside and cleanly rendering the money without messy artifacting."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Turn the man's head to face straight forward toward the background, perfectly matching the mannequin's original head direction.\n- Straighten the man's left arm to hang straight down at his side, holding the envelope next to his leg without bending the elbow.\n- Reconstruct the left hand and envelope to have correct human anatomy, properly positioning the thumb on the inside and cleanly rendering the money without messy artifacting.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh8__bgfirst_bg.png",
   "bg_asset_id": "aa5ae64e-3e09-4af0-9296-87fac507ce82",
   "bg_record_key": "S7sh8::bgfirst_bg",
   "chain_winner": true,
   "authority": "lane_conti_only"
  },
  "ref_mode": "lane(map_marker): 스케치+prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S7sh6"
  }
 },
 "S8sh1": {
  "input_fingerprint": "2cad6554cfdd5601",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 편의점 앞 벤치에서 김지후(강수리영)가 정인우의 손에 쥐어진 두툼한 돈봉투의 끝을 강하게 움켜쥐고 자기 쪽으로 팽팽하게 끌어당기는 mid-action 순간.\n\nLOCATION (lock): At the nighttime bench directly in front of the convenience store. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The same thick money envelope remains between them as Jihu pulls it from Inwoo; Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 3,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "이전 샷의 배경을 그대로 유지하면서 지후와 인우(팔 부분)가 돈봉투를 두고 팽팽히 당기는 액션을 해부학적 오류 없이 완벽하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "지후의 손이 돈봉투와 물리적으로 불가능하게 융합되어 해부학적 규칙을 심각하게 위반했습니다(Hard Violation)."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "이전 샷에 존재하지 않던 다수의 포스터가 배경에 창조되었고, 봉투를 쥔 손의 형태와 구조가 왜곡되었습니다(Hard Violation)."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S6sh7_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:755998>"
   },
   {
    "label": "PROP REFERENCE — 두툼한 돈봉투: the exact object appearing in this shot; match its look, material and wear exactly.",
    "path": "<bytes:1264547>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지후가 돈봉투를 움켜쥔 오른손의 해부학적 구조가 잘못되었습니다 (오른팔에 왼손이 붙어 있는 것처럼 엄지손가락이 반대쪽인 바깥쪽에 위치함).",
     "fix_en": "Redraw Jihu's gripping right hand to have correct human anatomy, ensuring the thumb is positioned on the side closest to her body.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "돈봉투가 레퍼런스의 각진 직육면체 형태와 테이프 디테일이 아닌, 단순한 원통형의 구겨진 종이봉투로 다르게 묘사되었습니다.",
     "fix_en": "Reshape the money envelope between their hands to exactly match the rigid, blocky rectangular form and taped details shown in the reference prop image.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "인우의 블레이저 소매 끝에 레퍼런스 의상(반팔 또는 소매가 보이지 않는 이너)에는 없는 어두운 색의 긴 소매가 튀어나와 있습니다.",
     "fix_en": "Remove the dark inner sleeve cuff protruding from Inwoo's blazer, showing only his bare wrist extending directly from the navy blazer sleeve.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw Jihu's gripping right hand to have correct human anatomy, ensuring the thumb is positioned on the side closest to her body.\n- Reshape the money envelope between their hands to exactly match the rigid, blocky rectangular form and taped details shown in the reference prop image.\n- Remove the dark inner sleeve cuff protruding from Inwoo's blazer, showing only his bare wrist extending directly from the navy blazer sleeve.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S6sh7"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S8sh5": {
  "input_fingerprint": "f030cd02c55d7a51",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 기묘한 기시감과 설명할 수 없는 서늘함에 사로잡혀 창백하게 굳어버린 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): On the bench outside the convenience store at night. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the nighttime convenience-store frontage, bench area, and exterior lighting. Exclude the envelope tugging action and use no unnecessary bystanders.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu retains the thick money envelope and still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 5,
   "A": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "샷 텍스트에 명시된 지후의 창백한 얼굴을 중심으로 클로즈업을 잘 구현했으며, 불필요한 인물을 배제하라는 지시를 완벽히 따랐습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "주요 소품과 표정은 양호하나, 샷 텍스트에 없는 인물을 배경에 추가하여 '불필요한 구경꾼 배제' 지침을 위반했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "지시문에 없는 인물의 어깨를 걸고 찍는 앵글을 사용하여 단독 얼굴 클로즈업이라는 핵심 연출을 크게 훼손했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S8sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1092285>"
   },
   {
    "label": "PROP REFERENCE — 두툼한 돈봉투: the exact object appearing in this shot; match its look, material and wear exactly.",
    "path": "<bytes:1264547>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "목걸이 펜던트가 레퍼런스의 화살촉 모양 뼈 장식이 아닌 날개 달린 형태로 잘못 생성되었습니다.",
     "fix_en": "Replace the winged pendant with the simple, arrowhead-shaped carved bone pendant from the reference still.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "돈봉투가 납작하게 표현되었으며, 레퍼런스에 있는 지폐 뭉치가 보이는 창과 수직 접은 선 디테일이 없습니다.",
     "fix_en": "Modify the envelope to be a bulky paper bag with a vertical center seam and a cut-out window revealing a stack of 10,000 won bills, exactly matching the prop reference.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "샷 텍스트에서 '얼굴 클로즈업'을 요구했으나, 상반신과 손까지 모두 포함된 넓은 샷(미디엄 클로즈업)으로 연출되었습니다.",
     "fix_en": "Camera must be moved significantly closer to frame a close-up restricted to Jihu's face, excluding her hands, lap, and lower torso.",
     "needs_regeneration": false,
     "unfixable": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the winged pendant with the simple, arrowhead-shaped carved bone pendant from the reference still.\n- Modify the envelope to be a bulky paper bag with a vertical center seam and a cut-out window revealing a stack of 10,000 won bills, exactly matching the prop reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S8sh1"
  }
 },
 "S9sh3::bgfirst_bg": {
  "input_fingerprint": "1fd52108458ff0b1",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 소파 쪽을 바라보며 얼굴이 하얗게 질린 채 경악으로 굳어버린 강민숙(홍은미)의 클로즈업.\n\nLOCATION (lock): Inside the compact living room of a two-room multiplex home, facing the sofa from the entry.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the living-room side of the doorway at 강민숙(홍은미)'s eye level, finish the dolly-in at close range with her face slightly off-center and the doorway edge compressing one side of the frame. The emphasized shift is distance: her trembling stare remains aimed toward the unseen sofa as her face loses color and her breath catches.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 강민숙(홍은미) in the middle-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 거실 쪽 현관 경계 (민숙이 거실로 들어서다 멈춘 위치) — The living-room-facing edge appears beside 강민숙(홍은미), with the entrance space behind her; used as Creates a hard vertical frame beside her face and preserves the entrance axis; 소파 (흰 머리 여자가 앉아 있음) — Positioned deeper inside the living room along her off-screen eyeline; used as Remains outside the tight composition as the source of her fixed attention.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 소파 쪽을 바라보며 얼굴이 하얗게 질린 채 경악으로 굳어버린 강민숙(홍은미)의 클로즈업.\n\nLOCATION (lock): Inside the compact living room of a two-room multiplex home, facing the sofa from the entry.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the living-room side of the doorway at 강민숙(홍은미)'s eye level, finish the dolly-in at close range with her face slightly off-center and the doorway edge compressing one side of the frame. The emphasized shift is distance: her trembling stare remains aimed toward the unseen sofa as her face loses color and her breath catches.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 강민숙(홍은미) in the middle-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 거실 쪽 현관 경계 (민숙이 거실로 들어서다 멈춘 위치) — The living-room-facing edge appears beside 강민숙(홍은미), with the entrance space behind her; used as Creates a hard vertical frame beside her face and preserves the entrance axis; 소파 (흰 머리 여자가 앉아 있음) — Positioned deeper inside the living room along her off-screen eyeline; used as Remains outside the tight composition as the source of her fixed attention.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S9sh3__bgfirst_bg.png",
  "asset_id": "fbfe6251-1cbd-40f6-b6f7-50286426ff30",
  "input_asset_ids": [
   "05c063ab-bb47-4d17-ab82-45e4cd749bfb",
   "41400a21-5b9d-4d6d-bfa4-ebb3bff48a6f"
  ]
 },
 "S9sh3": {
  "input_fingerprint": "f405de1c0df1409c",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 소파 쪽을 바라보며 얼굴이 하얗게 질린 채 경악으로 굳어버린 강민숙(홍은미)의 클로즈업.\n\nLOCATION (lock): Inside the compact living room of a two-room multiplex home, facing the sofa from the entry. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 소파 쪽을 바라보며 얼굴이 하얗게 질린 채 경악으로 굳어버린 강민숙(홍은미)의 클로즈업.\n\nLOCATION (lock): Inside the compact living room of a two-room multiplex home, facing the sofa from the entry. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 소파 쪽을 바라보며 얼굴이 하얗게 질린 채 경악으로 굳어버린 강민숙(홍은미)의 클로즈업.\n\nLOCATION (lock): Inside the compact living room of a two-room multiplex home, facing the sofa from the entry. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S9sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S9sh3.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L08B02.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "화살표 마커가 그대로 노출되었고 프롬프트에 없는 인물이 배경에 추가되어 심각한 규정 위반입니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시된 위치에서 마커나 추가 인물 없이, 창백하게 질려 경악하는 인물의 클로즈업을 사실적으로 잘 구현했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "화살표 마커가 그대로 노출되었고 프롬프트에 없는 인물이 배경에 추가되어 심각한 규정 위반입니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시된 위치에서 마커나 추가 인물 없이, 창백하게 질려 경악하는 인물의 클로즈업을 사실적으로 잘 구현했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "치명적 위반이 없으며 클로즈업 샷과 경악으로 굳은 표정을 잘 표현했으나, 카메라 방향이 소파가 아닌 현관 쪽을 향한 점이 아쉽습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "화면에 눈에서 뻗어나가는 지시선(화살표)이 그대로 노출되었고, 지문에 없는 인물이 소파에 추가되어 치명적 위반으로 실격입니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "치명적 위반이 없으며 클로즈업 샷과 경악으로 굳은 표정을 잘 표현했으나, 카메라 방향이 소파가 아닌 현관 쪽을 향한 점이 아쉽습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "화면에 눈에서 뻗어나가는 지시선(화살표)이 그대로 노출되었고, 지문에 없는 인물이 소파에 추가되어 치명적 위반으로 실격입니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 5,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 5,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 2,
    "verdict_ko": "화살표 마커가 그대로 노출되었고 프롬프트에 없는 인물이 배경에 추가되어 심각한 규정 위반입니다."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지시된 위치에서 마커나 추가 인물 없이, 창백하게 질려 경악하는 인물의 클로즈업을 사실적으로 잘 구현했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L08B02.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "인물이 소파(왼쪽) 쪽을 바라보지 않고 반대편(오른쪽)을 바라보고 있습니다.",
     "fix_en": "Turn the woman's head and eyes to her right (the viewer's left) so she is looking directly at the sofa.",
     "unfixable": false
    },
    {
     "issue_ko": "카메라는 현관에서 소파 방향을 향해야 하지만, 현재 거실 안쪽에서 현관 쪽을 바라보는 구도입니다.",
     "fix_en": "The camera is facing the entry door instead of facing the sofa from the entry; this viewpoint error cannot be fixed in-place.",
     "unfixable": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Turn the woman's head and eyes to her right (the viewer's left) so she is looking directly at the sofa.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S9sh3__bgfirst_bg.png",
   "bg_asset_id": "fbfe6251-1cbd-40f6-b6f7-50286426ff30",
   "bg_record_key": "S9sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S9sh6": {
  "input_fingerprint": "de0c785a903db941",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 현관문을 꽉 막고 선 거대한 덩치의 고산들 가슴팍 앞에서 뒤로 넘어질 듯 몸이 젖혀진 강민숙(홍은미)의 상태.\n\nLOCATION (lock): At the interior entryway of the multiplex home, directly before the blocked front door. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the modest living room, sofa, entrance door, and night interior lighting. Exclude the seated white-haired woman's earlier composition and keep the doorway blocked by the large man.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 5,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "레퍼런스의 공간 구조를 유지하면서 거대한 남자의 가슴 앞에서 뒤로 넘어질 듯한 여성의 자세를 가장 정확하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "공간과 인물들의 외형은 유지했으나, 여성이 여전히 소파에 앉아 있어 '뒤로 넘어질 듯 몸이 젖혀진' 행동 지시를 위반했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "인물들의 동작은 묘사했으나, 현관과 거실의 위치가 완전히 뒤바뀌는 등 레퍼런스의 공간 구조를 무시한 심각한 오류가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S9sh3_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "기준 이미지의 우측에 있던 현관이 좌측으로 이동했고, 복도와 인터폰이 두 군데로 중복 생성되는 등 배경 구조가 완전히 왜곡되었습니다.",
     "fix_en": "Rebuild the background architecture to match the reference exactly, with a single hallway and the step-down entrance positioned on the right.",
     "needs_regeneration": true,
     "unfixable": true
    },
    {
     "issue_ko": "인물들이 현관문 바로 앞이 아닌 거실 한가운데에 배치되어 있어, 남자가 현관을 막고 있지 않고 실제 현관문이 좌측 배경에 텅 빈 채로 보입니다.",
     "fix_en": "Relocate the camera and characters directly to the front entrance, ensuring the large man's body completely fills and blocks the doorway right behind the woman.",
     "needs_regeneration": false,
     "unfixable": true
    }
   ]
  },
  "fix_skipped": true,
  "fix_skip_reason": "all_issues_unfixable",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S9sh3"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S10sh3": {
  "input_fingerprint": "812f3cb757cd46a8",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 들고 있는 핸드폰 화면에 '도망쳐'라는 짧은 메시지가 선명하게 뜬 찰나.\n\nLOCATION (lock): Behind the counter on the convenience-store sales floor at night, under the store's active lighting. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still has the thick money envelope and wears her antler-bone pendant.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 3,
   "C": 2
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지후가 핸드폰을 들고 있는 모습과 화면의 메시지, 자연스러운 시선 방향과 소품의 위치를 정확하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "주인공이 핸드폰을 들고 있지 않거나 엉뚱한 인물이 등장했으며, 폰 화면이 카메라를 향해 부자연스럽게 노출되었습니다."
   },
   {
    "label": "C",
    "score": 2,
    "verdict_ko": "인물이 핸드폰을 들고 있다는 핵심 지시를 무시했으며, 허공에 떠 있는 텍스트는 명백한 규정 위반입니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L10B02.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "핸드폰 화면이 카메라 쪽을 향하고 있어, 화면을 보고 있는 지후의 시선 방향과 물리적으로 일치하지 않습니다.",
     "fix_en": "Rotate the phone so its screen directly faces Jihu's eyes.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "핸드폰 왼쪽을 잡고 있는 팔이 지후의 몸과 연결되지 않고 화면 좌측 하단에서 뻗어 나오는 해부학적 오류가 있습니다.",
     "fix_en": "Remove the arm and hand holding the left side of the phone, leaving Jihu holding the phone with only her right hand.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "지후의 유니폼 명찰에 프롬프트에서 지시하지 않은 임의의 문자가 포함되어 있습니다.",
     "fix_en": "Remove all text from the name tag on the uniform.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Rotate the phone so its screen directly faces Jihu's eyes.\n- Remove the arm and hand holding the left side of the phone, leaving Jihu holding the phone with only her right hand.\n- Remove all text from the name tag on the uniform.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S10sh5": {
  "input_fingerprint": "71badb5357f8785a",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 초조한 표정으로 핸드폰을 귀에 바짝 댄 채 어두운 편의점 입구 쪽으로 시선을 향한 김지후(강수리영)의 측면.\n\nLOCATION (lock): At the convenience-store counter, facing the dim entrance while the sales floor remains lit. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the dim convenience-store counter area, entrance sightline, and night lighting. Exclude the readable warning-message insert and keep only the phone call pose.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still has the money envelope and antler-bone pendant while holding her phone during the unanswered call.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 4,
   "A": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "측면 앵글, 초조한 표정, 통화 동작 및 레퍼런스의 세트 구성과 소품 유지가 지시사항과 가장 정확히 일치합니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "인물 연출은 무난하나, 레퍼런스와 달리 계산대와 출입문의 위치가 반대로 뒤집힌 심각한 공간 배치 오류가 있습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "양쪽 귀에 각각 다른 핸드폰을 쥐고 대고 있는 신체 및 사물 중복 생성 오류(Hard Violation)가 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S10sh3_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "통화 중인 스마트폰의 화면이 귀에 밀착되지 않고 바깥쪽을 향하고 있어 방향이 잘못되었습니다.",
     "fix_en": "Flip the smartphone in her hand so the screen faces inward against her ear, leaving the back of the phone facing outward.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "기준 사진에서는 카키색 재킷 위에 보라색/녹색 유니폼 조끼를 겉옷으로 입고 있으나, 생성된 사진은 조끼 위에 재킷을 덧입어 옷의 레이어링이 반대로 되었습니다.",
     "fix_en": "Change the clothing layering so the purple and green uniform vest is the outermost layer worn over the olive jacket.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Flip the smartphone in her hand so the screen faces inward against her ear, leaving the back of the phone facing outward.\n- Change the clothing layering so the purple and green uniform vest is the outermost layer worn over the olive jacket.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S10sh3"
  }
 },
 "S11sh2": {
  "input_fingerprint": "dfad06f9e544c41f",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 붉은 경광등이 켜진 경찰차가 김지후(강수리영) 앞 골목길에 급정거하여 차체가 앞쪽으로 약간 쏠려 있는 찰나.\n\nLOCATION (lock): On a narrow nighttime lane between multiplex homes, where a police car stops ahead. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu continues to carry her phone in hand and still wears her antler-bone pendant.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 9,
   "A": 6,
   "C": 5
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 9,
    "verdict_ko": "지후의 1인칭 시점(POV)을 채택하여 '지후 앞'이라는 공간적 위치를 영리하게 풀었으며, 급정거로 인해 차체가 앞으로 쏠리는 역동적인 찰나를 가장 완벽하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "지후의 소품(스마트폰, 뿔 목걸이)은 지문에 맞게 묘사되었으나, 경찰차가 인물 뒤에 배치되어 '지후 앞 골목길'이라는 위치 관계를 위반했고 급정거의 느낌도 부족합니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "지후를 프레임 안쪽 멀리 배치하여 경찰차와의 관계는 성립시켰으나, 차체가 앞으로 쏠리는 텍스트의 핵심적인 물리적 디테일이 시각적으로 전혀 드러나지 않습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L07B01.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "경찰차 앞부분이 급정거로 인해 아래로 쏠린(노즈 다운) 모습이 표현되지 않았습니다.",
     "fix_en": "Tilt the police car's body downward at the front to visually show hard braking.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "'no text' 지시를 위반하여 왼쪽 벽면의 어두운 표지판에 의미 없는 문자가 생성되었습니다.",
     "fix_en": "Remove all the gibberish text from the dark rectangular sign on the left wall.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Tilt the police car's body downward at the front to visually show hard braking.\n- Remove all the gibberish text from the dark rectangular sign on the left wall.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S11sh3": {
  "input_fingerprint": "c1a9f6b1749ee1ad",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 열린 운전석 문 밖으로 다급하고 굳은 표정의 얼굴을 내민 채 몸을 기울여 고정된 정인우의 모습.\n\nLOCATION (lock): Beside the open driver's door of a police car stopped in the multiplex-housing alley. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the narrow night alley, red emergency light, stopped car, and open driver-side area. Exclude the earlier empty exterior emphasis and show the driver leaning out.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still carries her phone and wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 4,
   "B": 3,
   "C": 7
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "프롬프트에 유일하게 허용된 인물(김지후) 지침을 어기고 임의의 남성을 생성하여 '발명된 인물'에 해당하는 중대한 위반을 범함."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "허용되지 않은 인물을 생성한 위반과 더불어, 차량 문 쪽에 정체불명의 디스플레이 장치가 떠 있는 물리적 배치 오류가 발생함."
   },
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "샷 텍스트의 이름에도 불구하고 인물 제한 지침을 준수하여 김지후를 정확히 묘사했으며, 야간 배경과 펜던트 소지 상태를 충실히 구현함."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S11sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1225321>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "차량의 색상이 참조 이미지(흰색 바탕에 파란색/노란색 경찰 도색 및 데칼)와 달리 완전히 어두운 색으로 생성되었습니다.",
     "fix_en": "Repaint the car body to match the reference: white with blue and yellow police stripes and decals.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "인물이 몸을 내민 열린 창문 공간을 가로지르는 물리적으로 불가능한 대각선 프레임(구조물)이 있습니다.",
     "fix_en": "Remove the diagonal metal bar crossing the open window space so the area the character leans through is completely empty.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Repaint the car body to match the reference: white with blue and yellow police stripes and decals.\n- Remove the diagonal metal bar crossing the open window space so the area the character leans through is completely empty.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S11sh2"
  }
 },
 "S12sh2::bgfirst_bg": {
  "input_fingerprint": "af39bc84259f4d1b",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 부서진 가구와 핏자국으로 난장판이 된 거실 한가운데에 무심하게 서 있는 고산들의 모습.\n\nLOCATION (lock): In the wrecked living room of the multiplex home, amid broken furniture and visible bloodstains.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the entrance at eye level, complete the revealing pan in a wide view across the wrecked living room and settle on 고산들 at the center amid broken furniture, objects, and visible bloodstains. Keep his full figure comparatively still and modest in scale, with the picture frame held loosely as the final visual stop before the camera advances into the room.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 부서진 가구와 깨진 물건들 (거실 곳곳에 흩어져 있음); used as Builds the primary field of destruction around the centered figure; 깨진 등 (깨져 있음) — Its damaged body is visible among the disordered room elements; used as Adds a specifically described broken vertical element within the wreckage; 화분 (난장판 속에 놓여 있음) — Seen among the scattered interior objects; used as Provides another explicit piece of the household destruction; 핏자국 (집안 여기저기에 선명하게 남아 있음); used as Marks the room's violence without assigning it to a particular victim; 사진액자 (고산들이 보고 있던 액자) — Its content-bearing front face angles toward 고산들 and remains unreadable to the entrance camera; used as Connects 고산들 to the object he had been examining.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 부서진 가구와 핏자국으로 난장판이 된 거실 한가운데에 무심하게 서 있는 고산들의 모습.\n\nLOCATION (lock): In the wrecked living room of the multiplex home, amid broken furniture and visible bloodstains.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the entrance at eye level, complete the revealing pan in a wide view across the wrecked living room and settle on 고산들 at the center amid broken furniture, objects, and visible bloodstains. Keep his full figure comparatively still and modest in scale, with the picture frame held loosely as the final visual stop before the camera advances into the room.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 부서진 가구와 깨진 물건들 (거실 곳곳에 흩어져 있음); used as Builds the primary field of destruction around the centered figure; 깨진 등 (깨져 있음) — Its damaged body is visible among the disordered room elements; used as Adds a specifically described broken vertical element within the wreckage; 화분 (난장판 속에 놓여 있음) — Seen among the scattered interior objects; used as Provides another explicit piece of the household destruction; 핏자국 (집안 여기저기에 선명하게 남아 있음); used as Marks the room's violence without assigning it to a particular victim; 사진액자 (고산들이 보고 있던 액자) — Its content-bearing front face angles toward 고산들 and remains unreadable to the entrance camera; used as Connects 고산들 to the object he had been examining.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S12sh2__bgfirst_bg.png",
  "asset_id": "58360bc0-450c-46a7-9dc3-6ff771f6b221",
  "input_asset_ids": [
   "ee932bba-7dbb-44f2-8b33-7edacfef5212",
   "a91eefed-67f0-4707-975f-77cb87e6f81f",
   "d74f097c-288e-4fa6-ac33-f9284505f0ba"
  ]
 },
 "S12sh2": {
  "input_fingerprint": "e945f9473d9cf53e",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 부서진 가구와 핏자국으로 난장판이 된 거실 한가운데에 무심하게 서 있는 고산들의 모습.\n\nLOCATION (lock): In the wrecked living room of the multiplex home, amid broken furniture and visible bloodstains. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The living room remains wrecked, with broken furniture, a shattered light and flowerpot, and conspicuous bloodstains. Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 부서진 가구와 핏자국으로 난장판이 된 거실 한가운데에 무심하게 서 있는 고산들의 모습.\n\nLOCATION (lock): In the wrecked living room of the multiplex home, amid broken furniture and visible bloodstains. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The living room remains wrecked, with broken furniture, a shattered light and flowerpot, and conspicuous bloodstains. Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 부서진 가구와 핏자국으로 난장판이 된 거실 한가운데에 무심하게 서 있는 고산들의 모습.\n\nLOCATION (lock): In the wrecked living room of the multiplex home, amid broken furniture and visible bloodstains. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The living room remains wrecked, with broken furniture, a shattered light and flowerpot, and conspicuous bloodstains. Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S12sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S12sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1225321>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L11B02.png"
    },
    {
     "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_multifamily_residence_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1225321>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시문대로 난장판이 된 거실 한가운데에 무심하게 서 있는 모습과 요구된 사슴뼈 펜던트를 매우 정확하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "지시문에 없는 액자를 들고 있어 '무심하게 서 있는' 연출에서 벗어났으며, 펜던트의 형태도 명확하지 않습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시문대로 난장판이 된 거실 한가운데에 무심하게 서 있는 모습과 요구된 사슴뼈 펜던트를 매우 정확하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "지시문에 없는 액자를 들고 있어 '무심하게 서 있는' 연출에서 벗어났으며, 펜던트의 형태도 명확하지 않습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 9,
      "verdict_ko": "거실 한가운데 무심하게 서 있는 인물의 모습과 뼈 펜던트 등 지시문의 요구사항과 캐릭터 외모를 매우 정확하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "지시문에 없는 액자와 정체불명의 연기가 추가되었으며, 금지된 레퍼런스의 카메라 구도 복사(문틀 포함)를 범하여 감점되었습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "거실 한가운데 무심하게 서 있는 인물의 모습과 뼈 펜던트 등 지시문의 요구사항과 캐릭터 외모를 매우 정확하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지시문에 없는 액자와 정체불명의 연기가 추가되었으며, 금지된 레퍼런스의 카메라 구도 복사(문틀 포함)를 범하여 감점되었습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 7,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 7,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지시문대로 난장판이 된 거실 한가운데에 무심하게 서 있는 모습과 요구된 사슴뼈 펜던트를 매우 정확하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "지시문에 없는 액자를 들고 있어 '무심하게 서 있는' 연출에서 벗어났으며, 펜던트의 형태도 명확하지 않습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L11B02.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_multifamily_residence_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1225321>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "로케이션 레퍼런스 사진의 카메라 구도와 실내 배치를 그대로 복사하여 프레이밍 지침을 위반했습니다.",
     "fix_en": "Restatement: The camera framing exactly duplicates the reference location photograph and cannot be fixed without moving the camera.",
     "needs_regeneration": false,
     "unfixable": true
    }
   ]
  },
  "fix_skipped": true,
  "fix_skip_reason": "all_issues_unfixable",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S12sh2__bgfirst_bg.png",
   "bg_asset_id": "58360bc0-450c-46a7-9dc3-6ff771f6b221",
   "bg_record_key": "S12sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate",
   "seed_attached": true
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S12sh10": {
  "input_fingerprint": "ab8196f2f5e6cacf",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 바닥에 쓰러진 김지후(강수리영)의 어깨를 고산들의 억센 손이 강하게 짓눌러 지후의 어깨가 바닥에 완전히 밀착된 정지 자세.\n\nLOCATION (lock): On the floor of the devastated multiplex-home living room. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the wrecked living room, broken furniture, bloodstains, and harsh night interior look. Exclude the man's earlier standing pose and retain only the required restraint action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The room remains wrecked and bloodstained; Jihu still wears her antler-bone pendant as she lies pinned on the floor.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 5,
   "A": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "억센 손이 어깨를 강하게 짓눌러 바닥에 밀착시킨 핵심 행동과 인물, 소품, 배경을 가장 정확하게 구현했습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "누군가 짓누르고 있으나 손이 어깨가 아닌 가슴(멱살) 부위를 쥐고 있어 지문의 구체적인 동작 묘사와 어긋납니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "어깨를 짓누르는 손이 전혀 등장하지 않으며, 인물이 스스로 상체를 지탱하고 있어 지문의 핵심 상황을 완전히 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S12sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "어깨를 누르는 남자의 손에 손가락이 6개(엄지 포함)로 잘못 생성되었습니다.",
     "fix_en": "Remove the extra index finger resting on the shoulder seam to restore a normal hand with exactly five digits.",
     "unfixable": false
    },
    {
     "issue_ko": "목걸이 펜던트가 참조 이미지의 조각된 화살촉 모양이 아닌 미니어처 동물 해골과 뿔 형태로 잘못 생성되었습니다.",
     "fix_en": "Replace the miniature skull and antlers pendant on the floor with the simple, carved bone arrowhead shape shown in the character reference.",
     "unfixable": false
    },
    {
     "issue_ko": "지후의 어깨가 바닥에 완전히 밀착되어야 한다는 지시문과 달리, 상체가 들려 있어 어깨가 공중에 떠 있습니다.",
     "fix_en": "Lower Jihu's shoulder, upper torso, and the pressing hand so that her shoulder is completely flattened against the floor.",
     "unfixable": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the extra index finger resting on the shoulder seam to restore a normal hand with exactly five digits.\n- Replace the miniature skull and antlers pendant on the floor with the simple, carved bone arrowhead shape shown in the character reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S12sh2"
  }
 },
 "S13sh3::bgfirst_bg": {
  "input_fingerprint": "5ffdd670698280db",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 핸드폰을 귀에 댄 고산들이 고개를 살짝 돌린 채 차가운 눈빛으로 뒷좌석을 곁눈질하는 찰나.\n\nLOCATION (lock): In the driver's area of an older SUV speeding along a dark rural highway.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the front passenger area at 고산들's eye level, frame him in a medium three-quarter profile with the steering wheel low in the foreground and the phone held against his ear. The pan ends as his chin turns toward the rear, keeping his face on the left half of the frame and opening the right side toward the unseen back seat so that his sidelong gaze becomes the visual bridge to the next shot.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: steering wheel (held with one hand) — Its upper arc faces the camera below 고산들's driving hand; used as Provides a low foreground reference for his divided attention while driving; front seat backs and cabin gap (fixed in place) — The seat backs angle away from the camera, leaving the rearward gap visible beside 고산들; used as Creates the open visual corridor toward the next rear-seat reveal; SUV side window (closed) — The window is seen obliquely with the night road passing beyond it; used as Carries the moving night exterior at the edge of the composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 핸드폰을 귀에 댄 고산들이 고개를 살짝 돌린 채 차가운 눈빛으로 뒷좌석을 곁눈질하는 찰나.\n\nLOCATION (lock): In the driver's area of an older SUV speeding along a dark rural highway.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the front passenger area at 고산들's eye level, frame him in a medium three-quarter profile with the steering wheel low in the foreground and the phone held against his ear. The pan ends as his chin turns toward the rear, keeping his face on the left half of the frame and opening the right side toward the unseen back seat so that his sidelong gaze becomes the visual bridge to the next shot.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: steering wheel (held with one hand) — Its upper arc faces the camera below 고산들's driving hand; used as Provides a low foreground reference for his divided attention while driving; front seat backs and cabin gap (fixed in place) — The seat backs angle away from the camera, leaving the rearward gap visible beside 고산들; used as Creates the open visual corridor toward the next rear-seat reveal; SUV side window (closed) — The window is seen obliquely with the night road passing beyond it; used as Carries the moving night exterior at the edge of the composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S13sh3__bgfirst_bg.png",
  "asset_id": "ff059557-58cb-4429-9d4d-f5132dc9d334",
  "input_asset_ids": [
   "d48e58a1-b5e0-46e0-af61-6aa5eac4419c",
   "e6e6bba4-a085-49e7-a4eb-efee2c5c2fcd"
  ]
 },
 "S13sh3": {
  "input_fingerprint": "234f5d7ff3f05c47",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 핸드폰을 귀에 댄 고산들이 고개를 살짝 돌린 채 차가운 눈빛으로 뒷좌석을 곁눈질하는 찰나.\n\nLOCATION (lock): In the driver's area of an older SUV speeding along a dark rural highway. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul retains the framed graduation photograph taken from Minsuk’s wrecked home.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 핸드폰을 귀에 댄 고산들이 고개를 살짝 돌린 채 차가운 눈빛으로 뒷좌석을 곁눈질하는 찰나.\n\nLOCATION (lock): In the driver's area of an older SUV speeding along a dark rural highway. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul retains the framed graduation photograph taken from Minsuk’s wrecked home.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 핸드폰을 귀에 댄 고산들이 고개를 살짝 돌린 채 차가운 눈빛으로 뒷좌석을 곁눈질하는 찰나.\n\nLOCATION (lock): In the driver's area of an older SUV speeding along a dark rural highway. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul retains the framed graduation photograph taken from Minsuk’s wrecked home.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S13sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S13sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:876576>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L10B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:876576>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "휴대폰의 기능적 면(화면)을 귀 쪽으로 올바르게 향하게 들고 있으며, 지문이 요구한 짧은 머리와 뒷좌석을 곁눈질하는 차가운 눈빛 연기를 충실히 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "휴대폰의 액정 화면이 바깥을 향하도록 거꾸로 쥐고 있어 소품 방향 규칙을 위반한 심각한 오류가 있으며, 모자를 착용해 지문의 '짧게 깎은 검은 머리' 설정을 가렸습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "휴대폰의 기능적 면(화면)을 귀 쪽으로 올바르게 향하게 들고 있으며, 지문이 요구한 짧은 머리와 뒷좌석을 곁눈질하는 차가운 눈빛 연기를 충실히 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "휴대폰의 액정 화면이 바깥을 향하도록 거꾸로 쥐고 있어 소품 방향 규칙을 위반한 심각한 오류가 있으며, 모자를 착용해 지문의 '짧게 깎은 검은 머리' 설정을 가렸습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "운전석에 앉아 핸드폰을 귀에 댄 채 뒷좌석을 곁눈질하는 인물의 행동과 소품 방향이 지시문과 정확히 일치합니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "인물이 운전석이 아닌 조수석에 앉아 있으며, 핸드폰의 화면 방향이 잘못되어 감점되었습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "운전석에 앉아 핸드폰을 귀에 댄 채 뒷좌석을 곁눈질하는 인물의 행동과 소품 방향이 지시문과 정확히 일치합니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "인물이 운전석이 아닌 조수석에 앉아 있으며, 핸드폰의 화면 방향이 잘못되어 감점되었습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 7
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 15,
   "B": 7
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "휴대폰의 기능적 면(화면)을 귀 쪽으로 올바르게 향하게 들고 있으며, 지문이 요구한 짧은 머리와 뒷좌석을 곁눈질하는 차가운 눈빛 연기를 충실히 구현했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "휴대폰의 액정 화면이 바깥을 향하도록 거꾸로 쥐고 있어 소품 방향 규칙을 위반한 심각한 오류가 있으며, 모자를 착용해 지문의 '짧게 깎은 검은 머리' 설정을 가렸습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L10B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "캐릭터의 위치와 크기가 레이아웃 스케치와 일치하지 않으며, 너무 오른쪽으로 치우쳐 배치되어 스케치에서는 가려져야 할 운전석 헤드레스트가 드러나 보입니다.",
     "fix_en": "Scale up the character and move him to the left to match the layout sketch exactly, ensuring his head completely covers the driver's seat headrest."
    },
    {
     "issue_ko": "운전대를 잡은 왼손의 위치가 레이아웃 스케치(상단 우측)와 달리 너무 아래쪽(하단 우측)에 위치해 있습니다.",
     "fix_en": "Reposition the left hand to grip the upper-right section of the steering wheel exactly as depicted in the layout sketch."
    },
    {
     "issue_ko": "스티어링 휠 좌측 하단 테두리에 원본 배경 이미지에는 없는 회색 얼룩 형태의 AI 아티팩트가 생성되었습니다.",
     "fix_en": "Remove the greyish artifact on the bottom-left rim of the steering wheel to restore the exact appearance of the reference background."
    },
    {
     "issue_ko": "센터 콘솔에 놓인 액자 속 졸업 사진 인물의 얼굴이 심하게 일그러지고 뭉개져 있습니다.",
     "fix_en": "Repair the distorted face in the framed graduation photograph so it appears as a clear, undistorted human face."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Scale up the character and move him to the left to match the layout sketch exactly, ensuring his head completely covers the driver's seat headrest.\n- Reposition the left hand to grip the upper-right section of the steering wheel exactly as depicted in the layout sketch.\n- Remove the greyish artifact on the bottom-left rim of the steering wheel to restore the exact appearance of the reference background.\n- Repair the distorted face in the framed graduation photograph so it appears as a clear, undistorted human face.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S13sh3__bgfirst_bg.png",
   "bg_asset_id": "ff059557-58cb-4429-9d4d-f5132dc9d334",
   "bg_record_key": "S13sh3::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S13sh4": {
  "input_fingerprint": "57a25bf70bc49780",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 차량 뒷좌석에 눕혀진 채 손발이 묶이고 두 눈을 꾹 감고 있는 김지후(강수리영)의 모습.\n\nLOCATION (lock): Across the rear bench of the moving older SUV, behind the driver's seat. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the old SUV interior, dark road lighting, and rear-seat layout. Exclude the driver's phone-call pose and center the bound passenger on the rear seat.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is stretched out lengthwise across the SUV's rear seat with both wrists and ankles bound, her torso and head resting against the seat cushions.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains stretched out lengthwise across the SUV's rear seat with both wrists and ankles bound, her torso and head resting against the seat cushions, still wearing her antler-bone pendant. Sandeul retains the framed photograph.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 5,
   "C": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "이동 중인 차량의 모션 블러와 이전 샷의 뒷좌석 디테일(탄력 스트랩)을 정확히 유지했으며, 가로로 길게 누운 지정 포즈를 잘 구현했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "누운 포즈는 적절하나 이전 샷에 있던 좌석의 스트랩이 누락되었고, 창밖 풍경에 움직임이 없어 정지된 차량처럼 보입니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "눈을 꾹 감은 표정은 텍스트와 가장 잘 맞으나, 가로로 길게 누워야 하는 절대적 지정 포즈를 위반하고 상체를 웅크리고 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S13sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1252442>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "밧줄로 묶인 두 손의 해부학적 구조가 뭉개져, 위쪽 손 아래로 정체불명의 기형적인 손가락들이 여러 개 튀어나와 있습니다.",
     "fix_en": "Repair the bound hands to show exactly two anatomically correct hands, removing the extra jumbled fingers underneath."
    },
    {
     "issue_ko": "지문에서 '두 눈을 꾹 감고 있는' 상태를 요구했으나, 눈가에 긴장감이 전혀 없이 평온하게 수면 중인 것처럼 가볍게 눈을 감고 있습니다.",
     "fix_en": "Add tension to the character's eyelids and brow so her eyes appear tightly squeezed shut."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Repair the bound hands to show exactly two anatomically correct hands, removing the extra jumbled fingers underneath.\n- Add tension to the character's eyelids and brow so her eyes appear tightly squeezed shut.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S13sh3"
  }
 },
 "S14sh2::bgfirst_bg": {
  "input_fingerprint": "b878ebe7a64f48f1",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 앰뷸런스 뒷문턱에 걸터앉아 부러진 한쪽 팔을 팔걸이 밴드로 고정한 채 지친 표정을 짓고 있는 정인우의 모습.\n\nLOCATION (lock): At the open rear threshold of an ambulance parked in a multiplex-housing alley.\n\nTIME OF DAY (lock): evening.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At seated eye level beside the open rear of the ambulance, complete the dolly-in on 정인우 in a medium frame, placing him slightly off-center with the rear threshold and supported arm fully readable. Catch him perched rather than posed, shoulders sagging and his free hand stabilizing his weight as the shallow approach prepares to arc around his side.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: ambulance rear opening and threshold (open) — The open rear faces the camera, with the threshold running beneath 정인우; used as Frames 정인우's seated position and visually contains his injured posture; police vehicles (red and blue emergency lights flashing) — Their sides are visible deeper in the alley behind the ambulance; used as Supplies intermittent incident-scene color behind the subject; multi-family housing alley (police moving through the area and residents watching) — The alley recedes behind 정인우 toward the housing facade; used as Provides layered activity and the location toward which residents are watching.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 앰뷸런스 뒷문턱에 걸터앉아 부러진 한쪽 팔을 팔걸이 밴드로 고정한 채 지친 표정을 짓고 있는 정인우의 모습.\n\nLOCATION (lock): At the open rear threshold of an ambulance parked in a multiplex-housing alley.\n\nTIME OF DAY (lock): evening.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At seated eye level beside the open rear of the ambulance, complete the dolly-in on 정인우 in a medium frame, placing him slightly off-center with the rear threshold and supported arm fully readable. Catch him perched rather than posed, shoulders sagging and his free hand stabilizing his weight as the shallow approach prepares to arc around his side.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: ambulance rear opening and threshold (open) — The open rear faces the camera, with the threshold running beneath 정인우; used as Frames 정인우's seated position and visually contains his injured posture; police vehicles (red and blue emergency lights flashing) — Their sides are visible deeper in the alley behind the ambulance; used as Supplies intermittent incident-scene color behind the subject; multi-family housing alley (police moving through the area and residents watching) — The alley recedes behind 정인우 toward the housing facade; used as Provides layered activity and the location toward which residents are watching.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S14sh2__bgfirst_bg.png",
  "asset_id": "a14b6c63-330e-4c96-a344-d8e9df0f83e4",
  "input_asset_ids": [
   "1bcf5eac-7952-4056-ba6a-505df64347db",
   "8565fea3-097d-441f-855b-aced7cf3e131"
  ]
 },
 "S14sh2": {
  "input_fingerprint": "492d1d60032731c0",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): evening.\n\nSHOT TEXT (authoritative, Korean): 앰뷸런스 뒷문턱에 걸터앉아 부러진 한쪽 팔을 팔걸이 밴드로 고정한 채 지친 표정을 짓고 있는 정인우의 모습.\n\nLOCATION (lock): At the open rear threshold of an ambulance parked in a multiplex-housing alley. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains immobilized in a sling.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): evening.\n\nSHOT TEXT (authoritative, Korean): 앰뷸런스 뒷문턱에 걸터앉아 부러진 한쪽 팔을 팔걸이 밴드로 고정한 채 지친 표정을 짓고 있는 정인우의 모습.\n\nLOCATION (lock): At the open rear threshold of an ambulance parked in a multiplex-housing alley. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains immobilized in a sling.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): evening.\n\nSHOT TEXT (authoritative, Korean): 앰뷸런스 뒷문턱에 걸터앉아 부러진 한쪽 팔을 팔걸이 밴드로 고정한 채 지친 표정을 짓고 있는 정인우의 모습.\n\nLOCATION (lock): At the open rear threshold of an ambulance parked in a multiplex-housing alley. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains immobilized in a sling.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S14sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S14sh2.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L13B03.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지시문에 없는 다수의 군중과 경찰이 배경에 등장하여 '언급되지 않은 인물 배제' 규칙을 위반했습니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "구급차 문턱에 앉은 인물의 깁스 상태와 지친 표정, 다세대 주택 골목 배경 등 모든 지시사항을 정확히 구현했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지시문에 없는 다수의 군중과 경찰이 배경에 등장하여 '언급되지 않은 인물 배제' 규칙을 위반했습니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "구급차 문턱에 앉은 인물의 깁스 상태와 지친 표정, 다세대 주택 골목 배경 등 모든 지시사항을 정확히 구현했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 장소와 시간대를 완벽하게 구현했으며, 프롬프트에 명시된 인물만 정확히 묘사하여 지시사항을 잘 따랐습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "주인공의 포즈와 깁스 상태는 잘 표현되었으나, 프롬프트에 없는 다수의 구경꾼과 경찰관을 배경에 추가하여 '명시된 인물 외에는 등장하지 않는다'는 규칙을 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 장소와 시간대를 완벽하게 구현했으며, 프롬프트에 명시된 인물만 정확히 묘사하여 지시사항을 잘 따랐습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "주인공의 포즈와 깁스 상태는 잘 표현되었으나, 프롬프트에 없는 다수의 구경꾼과 경찰관을 배경에 추가하여 '명시된 인물 외에는 등장하지 않는다'는 규칙을 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "지시문에 없는 다수의 군중과 경찰이 배경에 등장하여 '언급되지 않은 인물 배제' 규칙을 위반했습니다."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "구급차 문턱에 앉은 인물의 깁스 상태와 지친 표정, 다세대 주택 골목 배경 등 모든 지시사항을 정확히 구현했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L13B03.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트에서 모든 텍스트를 금지했으나, 앰뷸런스 문 하단('119 구급대')과 창문, 그리고 왼쪽 건물 벽에 텍스트 및 숫자가 나타나 있습니다.",
     "fix_en": "Remove all text, numbers, and decals from the ambulance doors, windows, and the left background building wall.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "지정된 레퍼런스 사진은 탁 트인 넓은 도로와 스카이라인을 공간적 진실로 제시하지만, 생성된 이미지는 좁은 주택가 골목을 배경으로 하여 래퍼런스 장소와 완전히 불일치합니다.",
     "fix_en": "Replace the background alley and buildings with the wide multi-lane road, streetlights, and distant skyline environment shown in the reference location photograph.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all text, numbers, and decals from the ambulance doors, windows, and the left background building wall.\n- Replace the background alley and buildings with the wide multi-lane road, streetlights, and distant skyline environment shown in the reference location photograph.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S14sh2__bgfirst_bg.png",
   "bg_asset_id": "a14b6c63-330e-4c96-a344-d8e9df0f83e4",
   "bg_record_key": "S14sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S14sh3": {
  "input_fingerprint": "ccd3dd70778c22c6",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): evening.\n\nSHOT TEXT (authoritative, Korean): 앰뷸런스 쪽에 앉은 정인우를 향해 의구심 가득한 굳은 시선을 던지는 노 형사의 얼굴.\n\nLOCATION (lock): Beside the ambulance in the cordoned multiplex-housing alley, near the building entrance. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the evening alley, ambulance, flashing police lights, and injured man's seated position. Exclude the broader crowd emphasis and frame the detective's questioning look.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains supported by the sling.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 5,
   "C": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "인우를 향하는 형사의 시선 방향이 정확하게 연출되었으며, 인우의 기존 자세와 위치가 레퍼런스에 맞게 잘 유지되었습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "형사의 시선이 뒤에 있는 인우를 향하지 않고 엇갈리며, 고정되어야 할 인우의 고개 방향이 변형되었습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "형사의 시선이 인우와 완전히 반대를 향하고 있으며, 지문에 없는 다수의 경찰 엑스트라가 등장하여 지침을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S14sh2_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트에 텍스트 생성 금지 지시가 있으나, 배경의 건물 입구 위와 노란색 폴리스 라인에 문자가 나타나 있습니다.",
     "fix_en": "Remove all text and letters from the building entrance facade and the yellow police tape, leaving them as plain surfaces.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all text and letters from the building entrance facade and the yellow police tape, leaving them as plain surfaces.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S14sh2"
  }
 },
 "S15sh3::bgfirst_bg": {
  "input_fingerprint": "5c8ead9d2e290023",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 뒷좌석에 결박된 김지후(강수리영)가 두려움 가득한 눈빛으로 번쩍 눈을 뜬 채 앞쪽을 올려다보는 찰나.\n\nLOCATION (lock): In the rear-seat area of the parked older SUV near a dawn harbor.\n\nTIME OF DAY (lock): dawn.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the gap between the front seats at seat-back height, aim rearward and slightly downward in a medium frame, placing 김지후(강수리영)'s bound body diagonally across the rear bench while her suddenly opened face occupies the upper central area. Hold the front-seat edges as a narrow frame within the frame as she draws her shoulders inward and looks naturally upward toward 고산들 in the front.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: front seat backs (fixed in place) — Their rear-facing sides flank the image on left and right; used as Forms paired foreground borders that reveal 김지후(강수리영) through the center gap; rear bench seat (occupied) — Its seating surface extends laterally beneath 김지후(강수리영); used as Supports the diagonal placement of the bound figure; SUV windows (closed) — The closed windows reveal the pale brightening sky, dark sea, and moored fishing boats outside; used as Carries the explicitly described dawn waterfront beyond the cabin.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 뒷좌석에 결박된 김지후(강수리영)가 두려움 가득한 눈빛으로 번쩍 눈을 뜬 채 앞쪽을 올려다보는 찰나.\n\nLOCATION (lock): In the rear-seat area of the parked older SUV near a dawn harbor.\n\nTIME OF DAY (lock): dawn.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the gap between the front seats at seat-back height, aim rearward and slightly downward in a medium frame, placing 김지후(강수리영)'s bound body diagonally across the rear bench while her suddenly opened face occupies the upper central area. Hold the front-seat edges as a narrow frame within the frame as she draws her shoulders inward and looks naturally upward toward 고산들 in the front.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: front seat backs (fixed in place) — Their rear-facing sides flank the image on left and right; used as Forms paired foreground borders that reveal 김지후(강수리영) through the center gap; rear bench seat (occupied) — Its seating surface extends laterally beneath 김지후(강수리영); used as Supports the diagonal placement of the bound figure; SUV windows (closed) — The closed windows reveal the pale brightening sky, dark sea, and moored fishing boats outside; used as Carries the explicitly described dawn waterfront beyond the cabin.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S15sh3__bgfirst_bg.png",
  "asset_id": "98bb2c10-fc16-4668-bab0-dbb9e57f0c06",
  "input_asset_ids": [
   "712a04e5-b62e-48ed-a2ae-b5bf8f03b4b5",
   "2d12655d-1686-49bf-829c-9c263977cc24",
   "d74f097c-288e-4fa6-ac33-f9284505f0ba"
  ]
 },
 "S15sh3": {
  "input_fingerprint": "9a5434104a2d20c2",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 뒷좌석에 결박된 김지후(강수리영)가 두려움 가득한 눈빛으로 번쩍 눈을 뜬 채 앞쪽을 올려다보는 찰나.\n\nLOCATION (lock): In the rear-seat area of the parked older SUV near a dawn harbor. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is stretched out lengthwise across the SUV's rear seat with both wrists and ankles bound, her torso and head resting against the seat cushions.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains bound hand and foot in the rear seat, still wearing her antler-bone pendant. Sandeul retains the framed photograph.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 뒷좌석에 결박된 김지후(강수리영)가 두려움 가득한 눈빛으로 번쩍 눈을 뜬 채 앞쪽을 올려다보는 찰나.\n\nLOCATION (lock): In the rear-seat area of the parked older SUV near a dawn harbor. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is stretched out lengthwise across the SUV's rear seat with both wrists and ankles bound, her torso and head resting against the seat cushions.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains bound hand and foot in the rear seat, still wearing her antler-bone pendant. Sandeul retains the framed photograph.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 뒷좌석에 결박된 김지후(강수리영)가 두려움 가득한 눈빛으로 번쩍 눈을 뜬 채 앞쪽을 올려다보는 찰나.\n\nLOCATION (lock): In the rear-seat area of the parked older SUV near a dawn harbor. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is stretched out lengthwise across the SUV's rear seat with both wrists and ankles bound, her torso and head resting against the seat cushions.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains bound hand and foot in the rear seat, still wearing her antler-bone pendant. Sandeul retains the framed photograph.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S15sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S15sh3.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L11B01.png"
    },
    {
     "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_multifamily_residence_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 로케이션 사진의 배경과 구조물 레퍼런스를 완벽하게 반영하였으며, 결박된 채 앞을 올려다보는 인물의 자세와 표정도 지시문에 부합합니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "로케이션 사진이 공간의 기준임에도 창밖 풍경을 완전히 다른 항구 모습으로 변경하여 위치 조건(우선순위 3)을 위반했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 로케이션 사진의 배경과 구조물 레퍼런스를 완벽하게 반영하였으며, 결박된 채 앞을 올려다보는 인물의 자세와 표정도 지시문에 부합합니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "로케이션 사진이 공간의 기준임에도 창밖 풍경을 완전히 다른 항구 모습으로 변경하여 위치 조건(우선순위 3)을 위반했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "요구된 뼈 펜던트 소품을 정확히 착용하고 있으며, 차량 내부 환경과 결박되어 누운 자세의 지시를 충실히 따랐습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "차량 내부 묘사와 결박된 자세는 우수하나, 필수 요구 사항인 뿔뼈 펜던트 소품이 누락되었습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "요구된 뼈 펜던트 소품을 정확히 착용하고 있으며, 차량 내부 환경과 결박되어 누운 자세의 지시를 충실히 따랐습니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "차량 내부 묘사와 결박된 자세는 우수하나, 필수 요구 사항인 뿔뼈 펜던트 소품이 누락되었습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 12,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "지정된 로케이션 사진의 배경과 구조물 레퍼런스를 완벽하게 반영하였으며, 결박된 채 앞을 올려다보는 인물의 자세와 표정도 지시문에 부합합니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "로케이션 사진이 공간의 기준임에도 창밖 풍경을 완전히 다른 항구 모습으로 변경하여 위치 조건(우선순위 3)을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L11B01.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_multifamily_residence_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "캐노니컬 포즈 규정에 따라 머리가 시트에 닿아 있어야 하지만 허공에 떠 있습니다.",
     "fix_en": "Lower Jihu's head so the back of it rests fully against the seat cushion."
    },
    {
     "issue_ko": "유지되어야 할 소품인 뿔 펜던트가 지후의 목에 없습니다.",
     "fix_en": "Add an antler-bone pendant on a string around Jihu's neck."
    },
    {
     "issue_ko": "뒷유리창 밖 배경이 로케이션 사진(탁 트인 도로 뷰)과 달리 잘못된 형태의 거대한 건물로 막혀 있습니다.",
     "fix_en": "Replace the building seen through the rear window with the distant street view and traffic lights shown in the LOCATION photograph.",
     "needs_regeneration": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Lower Jihu's head so the back of it rests fully against the seat cushion.\n- Add an antler-bone pendant on a string around Jihu's neck.\n- Replace the building seen through the rear window with the distant street view and traffic lights shown in the LOCATION photograph.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S15sh3__bgfirst_bg.png",
   "bg_asset_id": "98bb2c10-fc16-4668-bab0-dbb9e57f0c06",
   "bg_record_key": "S15sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate",
   "seed_attached": true
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S15sh5": {
  "input_fingerprint": "78cfd9e7b0598ba7",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 운전석 룸미러를 통해 뒷좌석의 김지후(강수리영)를 말없이 응시하는 고산들의 무거운 시선.\n\nLOCATION (lock): At the driver's seat of the older SUV, with pale dawn and a dark harbor visible through the windows. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the dawn-lit SUV interior, bound passenger in the rear, and marina glimpses through the windows. Exclude her earlier direct upward gaze and use the rearview mirror as the viewing axis.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains bound in the rear seat and wears her antler-bone pendant; the framed photograph remains in Sandeul’s possession.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 4,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "룸미러를 통한 인물들의 시선과 배치, 지후의 결박 자세, 항구 배경을 물리적 오류 없이 정확하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "운전석이 비어있으나 룸미러에는 운전자가 비치는 물리적으로 불가능한 화면 구성(Hard Violation)이 발생했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "룸미러가 두 개로 중복 생성되었고 이전 샷의 지후 자세가 변경된 치명적인 오류(Hard Violation)가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S15sh3_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "전방을 향한 카메라 구도에서 룸미러가 뒷좌석과 운전자의 얼굴을 동시에 비추는 것은 광학적으로 불가능하며, 운전자의 얼굴이 거울 안 허공에 합성된 것처럼 떠 있습니다.",
     "fix_en": "Remove the driver's floating face from the left side of the rearview mirror and fill that area with the continued reflection of the back seat.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "거울에 비친 뒷좌석의 지후가 상체를 비스듬히 세우고 기대어 있어, 시트에 완전히 평평하게 누워 있는 레퍼런스의 자세와 다릅니다.",
     "fix_en": "Reposition the reflection of the woman so her body lies completely flat horizontally on the back seat, with her head resting down on the bottom seat cushion.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "지후의 손목에 있는 결박 도구가 레퍼런스의 두꺼운 검은색 수갑이 아닌 얇은 은색 금속 수갑으로 되어 있습니다.",
     "fix_en": "Change the silver metal handcuffs on the woman's wrists to thick, black heavy-duty restraints matching the reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Reposition the reflection of the woman so her body lies completely flat horizontally on the back seat, with her head resting down on the bottom seat cushion.\n- Change the silver metal handcuffs on the woman's wrists to thick, black heavy-duty restraints matching the reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S15sh3"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S16sh2::bgfirst_bg": {
  "input_fingerprint": "624e72d58ccf3fa9",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 모니터 불빛에 비친 정인우가 미간을 잔뜩 찌푸린 채 적개심 가득한 표정을 하고 있는 얼굴 클로즈업.\n\nLOCATION (lock): At a computer workstation inside a police detectives' office, illuminated by the monitor and office lights.\n\nTIME OF DAY (lock): dawn.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At 정인우's eye height and a front three-quarter angle, hold a close frame in which his furrowed brow, narrowed eyes, and tightened jaw occupy most of the image while a narrow monitor edge remains near his eyeline. His torso pitches toward the footage, with the sling only partly entering the lower frame so that the hostile expression remains primary.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 정인우 in the middle-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: computer monitor (on) — The active screen faces 정인우 and displays CCTV footage of 고산들 carrying the unconscious 김지후(강수리영) out of the house; only a narrow portion need remain visible to camera; used as Provides the explicitly stated illumination and the unseen object of 정인우's hostility; police-station workstation (in use) — The workstation recedes beside and behind 정인우; used as Places the close observation within the detectives' office without distracting from the face.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 모니터 불빛에 비친 정인우가 미간을 잔뜩 찌푸린 채 적개심 가득한 표정을 하고 있는 얼굴 클로즈업.\n\nLOCATION (lock): At a computer workstation inside a police detectives' office, illuminated by the monitor and office lights.\n\nTIME OF DAY (lock): dawn.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At 정인우's eye height and a front three-quarter angle, hold a close frame in which his furrowed brow, narrowed eyes, and tightened jaw occupy most of the image while a narrow monitor edge remains near his eyeline. His torso pitches toward the footage, with the sling only partly entering the lower frame so that the hostile expression remains primary.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 정인우 in the middle-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: computer monitor (on) — The active screen faces 정인우 and displays CCTV footage of 고산들 carrying the unconscious 김지후(강수리영) out of the house; only a narrow portion need remain visible to camera; used as Provides the explicitly stated illumination and the unseen object of 정인우's hostility; police-station workstation (in use) — The workstation recedes beside and behind 정인우; used as Places the close observation within the detectives' office without distracting from the face.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S16sh2__bgfirst_bg.png",
  "asset_id": "873d1997-d640-433b-8d43-9637dcdb1423",
  "input_asset_ids": [
   "1f8e3394-b798-45c3-b72e-8260cefda4bc",
   "7189c701-72bb-4732-be38-d662b554048f"
  ]
 },
 "S16sh2": {
  "input_fingerprint": "49c41a641df45939",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 모니터 불빛에 비친 정인우가 미간을 잔뜩 찌푸린 채 적개심 가득한 표정을 하고 있는 얼굴 클로즈업.\n\nLOCATION (lock): At a computer workstation inside a police detectives' office, illuminated by the monitor and office lights. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in its sling.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 모니터 불빛에 비친 정인우가 미간을 잔뜩 찌푸린 채 적개심 가득한 표정을 하고 있는 얼굴 클로즈업.\n\nLOCATION (lock): At a computer workstation inside a police detectives' office, illuminated by the monitor and office lights. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in its sling.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 모니터 불빛에 비친 정인우가 미간을 잔뜩 찌푸린 채 적개심 가득한 표정을 하고 있는 얼굴 클로즈업.\n\nLOCATION (lock): At a computer workstation inside a police detectives' office, illuminated by the monitor and office lights. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in its sling.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S16sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S16sh2.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L13B04.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "클로즈업 프레이밍, 모니터 조명, 창밖의 새벽 시간대와 적대적인 표정을 잘 살렸으나, 첨부된 차량 내부 레퍼런스 장소를 무시하고 텍스트(사무실)만 따랐습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "요구된 샷 구도와 팔걸이를 보여주지만, 표정의 적개심이 상대적으로 약하며 새벽 시간대와 레퍼런스의 차량 내부 장소 모두 반영되지 않았습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "클로즈업 프레이밍, 모니터 조명, 창밖의 새벽 시간대와 적대적인 표정을 잘 살렸으나, 첨부된 차량 내부 레퍼런스 장소를 무시하고 텍스트(사무실)만 따랐습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "요구된 샷 구도와 팔걸이를 보여주지만, 표정의 적개심이 상대적으로 약하며 새벽 시간대와 레퍼런스의 차량 내부 장소 모두 반영되지 않았습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시된 얼굴 클로즈업 앵글과 모니터 조명 아래 적개심 어린 표정을 잘 구현했으며, 언급되지 않은 인물을 배제하여 지침을 준수했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "표정과 앵글은 양호하나, 샷 텍스트에 명시되지 않은 배경 인물들을 임의로 추가하여 심각한 규정 위반(Invented people)이 발생했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시된 얼굴 클로즈업 앵글과 모니터 조명 아래 적개심 어린 표정을 잘 구현했으며, 언급되지 않은 인물을 배제하여 지침을 준수했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "표정과 앵글은 양호하나, 샷 텍스트에 명시되지 않은 배경 인물들을 임의로 추가하여 심각한 규정 위반(Invented people)이 발생했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 8,
     "B": 11
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 8,
   "B": 11
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "클로즈업 프레이밍, 모니터 조명, 창밖의 새벽 시간대와 적대적인 표정을 잘 살렸으나, 첨부된 차량 내부 레퍼런스 장소를 무시하고 텍스트(사무실)만 따랐습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "요구된 샷 구도와 팔걸이를 보여주지만, 표정의 적개심이 상대적으로 약하며 새벽 시간대와 레퍼런스의 차량 내부 장소 모두 반영되지 않았습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L13B04.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "배경이 레퍼런스 사진에 지정된 장소(차량 내부)와 일치하지 않고 사무실 공간으로 묘사되었습니다.",
     "fix_en": "Replace the office background with the car interior environment shown in the reference location photograph.",
     "needs_regeneration": false,
     "unfixable": true
    }
   ]
  },
  "fix_skipped": true,
  "fix_skip_reason": "all_issues_unfixable",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S16sh2__bgfirst_bg.png",
   "bg_asset_id": "873d1997-d640-433b-8d43-9637dcdb1423",
   "bg_record_key": "S16sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S16sh4": {
  "input_fingerprint": "bdb0fea1f407ad14",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 정인우가 손에 쥔 핸드폰 화면에 금월도 관련 문자 메시지가 떠 있는 상태의 클로즈업.\n\nLOCATION (lock): At the detectives' office workstation, where a phone is held beside the CCTV monitor. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the police-office workstation, monitor glow, and early-morning fluorescent ambience. Exclude the earlier hostile face close-up and emphasize only the phone message.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo still wears the arm sling while holding his phone.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 5,
   "A": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "지시된 팔걸이 착용 상태(왼팔 고정)를 정확히 유지하며 CCTV 앞 핸드폰 화면 클로즈업을 완벽히 연출함."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "핸드폰 화면과 CCTV 배경은 좋으나, 팔걸이를 한 왼팔까지 사용해 양손으로 쥐고 있어 부상 상태 유지 설정에 어긋남."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "팔걸이 끈이 완전히 누락되었으며, 레퍼런스에 없는 안경이 추가되어 인물 및 복장 일관성을 위반함."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S16sh2_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "스마트폰 화면이 카메라를 향하고 있어, 오른쪽의 인물은 화면이 아닌 기기의 뒷면을 보게 되는 모순된 방향입니다.",
     "fix_en": "Turn the mobile phone around so its screen faces the character on the right.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "재킷의 깃에 있는 앞 지퍼가 턱 아래(앞면)가 아닌 뒷목 부분에 위치해 있습니다.",
     "fix_en": "Move the jacket's collar zipper from the back of the neck to the front, under the chin.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "팔걸이 끈의 형태가 레퍼런스의 넓고 매끄러운 회색 밴드가 아닌 어두운 골지 무늬의 끈으로 바뀌었습니다.",
     "fix_en": "Change the arm sling strap to a wide, smooth, solid grey band matching the reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Move the jacket's collar zipper from the back of the neck to the front, under the chin.\n- Change the arm sling strap to a wide, smooth, solid grey band matching the reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S16sh2"
  }
 },
 "S17sh1": {
  "input_fingerprint": "caca65759639f97e",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 새파란 바다 위로 하얀 포말이 허공에 흩뿌려진 채, 뱃머리가 물살을 가르며 나아가는 순간의 5톤 낚시 어선 외부 전경.\n\nLOCATION (lock): On open blue water around a moving five-ton fishing boat, with white wake breaking from its bow. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 6,
   "C": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "'허공에 흩뿌려진 하얀 포말'이라는 핵심 묘사를 가장 역동적으로 구현했으며, 텍스트 금지 지시를 완벽히 준수했습니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "어선의 외부 전경은 잘 담아냈으나, 레퍼런스의 구도를 너무 비슷하게 차용했으며 포말이 허공에 흩뿌려지는 묘사가 약합니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "포말 묘사는 좋으나 선체에 글씨와 국기를 그려 넣어 'No text' 지시를 명백히 위반했고, 전경(wide shot)에서 너무 벗어났습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L14B01.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "레이더 스캐너와 조타실 창문 아래 외벽에 영문 텍스트가 적혀 있습니다.",
     "fix_en": "Remove the text from the radar scanner and the side of the cabin, leaving the surfaces completely blank.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the text from the radar scanner and the side of the cabin, leaving the surfaces completely blank.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S17sh6": {
  "input_fingerprint": "08a0894d523bd54b",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 통로 너머 뱃머리에 굳건히 서 있는 고산들의 뒷모습을 향해 시선을 고정한 김지후(강수리영)의 모습.\n\nLOCATION (lock): On the open passenger deck of the moving fishing boat, looking along its passage toward the bow. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the moving fishing boat, blue sea, foamy wake, and open daylight. Exclude the wide exterior-only framing and show the woman looking toward the man at the bow.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 3,
   "B": 9,
   "C": 6
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "프롬프트에 명시되지 않은 추가 인물들이 갑판에 등장하여 심각한 지침 위반입니다."
   },
   {
    "label": "B",
    "score": 9,
    "verdict_ko": "레퍼런스 선박의 좌현 구조와 구명환을 정확히 재현했으며, 뱃머리의 남자를 향한 시선과 뼈 펜던트 연출이 매우 충실합니다."
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "여성의 시선이 뱃머리의 남자가 아닌 측면을 향하고 있으며, 펜던트의 형태와 선박의 구조가 레퍼런스와 다소 차이가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S17sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "여자의 시선이 뱃머리에 서 있는 남자가 아닌 왼쪽 바다를 향해 있습니다.",
     "fix_en": "Turn the woman's head and eyes forward so she is looking directly at the man standing at the bow.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Turn the woman's head and eyes forward so she is looking directly at the man standing at the bow.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S17sh1"
  }
 },
 "S18sh1": {
  "input_fingerprint": "b4112916ba82cc5f",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 구형 SUV 조수석에 앉은 김지후(강수리영)가 두 손으로 고등학교 졸업식 사진 액자를 쥐고 내려다보는 모습.\n\nLOCATION (lock): In the front passenger seat of the older SUV near the dawn harbor. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu is now unbound, still wears her antler-bone pendant, and holds the framed graduation photograph taken from Minsuk’s home.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 8,
   "A": 6,
   "C": 2
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "지시문을 정확히 따르며, 조수석 구도와 졸업식 사진, 뼈 펜던트 및 의상 등 이전 샷과 레퍼런스의 디테일을 완벽하게 일치시켰습니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "전반적인 구도와 배경은 좋으나, 목걸이 펜던트의 재질과 디자인이 레퍼런스와 다르고 액자를 쥔 손의 묘사가 다소 어색합니다."
   },
   {
    "label": "C",
    "score": 2,
    "verdict_ko": "문자가 포함된 두 번째 액자가 가슴 쪽에 중복으로 생성되는 심각한 물리적/구조적 오류가 있어 규정 위반입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S15sh5_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "사진 액자의 앞면(기능적인 면)이 지후의 시선이 아닌 카메라 쪽을 향하고 있어, 인물이 액자의 옆면을 보게 되는 부자연스러운 방향입니다.",
     "fix_en": "Rotate the picture frame so the photograph directly faces Jihu's eyes, exposing the back of the frame to the camera.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Rotate the picture frame so the photograph directly faces Jihu's eyes, exposing the back of the frame to the camera.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S15sh5"
  }
 },
 "S18sh5": {
  "input_fingerprint": "f4a0e34f66120595",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 진지하고 단호한 눈빛으로 조수석의 김지후(강수리영)를 똑바로 쳐다보는 고산들의 굳은 표정.\n\nLOCATION (lock): Inside the older SUV's front cabin, between the driver's seat and the occupied passenger seat. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the old SUV cabin, dawn light, dashboard materials, and the photograph held by the passenger. Exclude the earlier downward-looking pose and frame the driver's firm gaze.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The framed graduation photograph remains with Jihu, who still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 4,
   "A": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "지시된 카메라 위치(두 좌석 사이)에서 운전자의 단호한 시선과 차량 내부 구도를 가장 정확하고 사실적으로 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "운전석 헤드레스트와 운전자의 신체 위치가 물리적으로 일치하지 않는 치명적인 구도 및 공간 배치 오류가 있습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "액자의 기능적 앞면이 카메라를 향하게 뒤집혀 있어 소품 방향 규칙을 명백히 위반한 결정적 오류가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S18sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지후의 목에 명시된 뿔뼈 펜던트 목걸이가 없습니다.",
     "fix_en": "Add the antler-bone pendant necklace over Jihu's white shirt.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the antler-bone pendant necklace over Jihu's white shirt.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S18sh1"
  }
 },
 "S19sh2": {
  "input_fingerprint": "f7024e7db920653f",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 친근하게 미소 지으며 얼굴을 가까이 들이민 김여상을 향해, 굳은 얼굴로 빤히 쳐다보는 김지후(강수리영)의 표정.\n\nLOCATION (lock): At a seat on the open rear deck of the moving fishing boat, slightly apart from the anglers. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 3,
   "C": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지문대로 여상을 향한 지후의 굳은 시선과 다가간 여상의 미소를 완벽히 연출했으며 프레이밍도 적절합니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "지후의 시선이 카메라를 향해 연출이 어긋났고, 이전 샷에 없던 추가 인물(낚시꾼)이 배경에 등장했습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "이전 샷의 배경 인물은 잘 유지했으나, 지후가 여상을 보지 않고 정면을 응시하여 핵심 텍스트를 놓쳤습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S17sh6_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "김지후의 목과 스웨터 깃 주변에 사슴뿔 펜던트의 목걸이 줄이 보이지 않습니다.",
     "fix_en": "Add the dark leather necklace cord around Jihu's neck, resting over her sweater collar.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the dark leather necklace cord around Jihu's neck, resting over her sweater collar.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S17sh6"
  }
 },
 "S19sh7": {
  "input_fingerprint": "33dd6d2144b1c770",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 고산들이 김지후(강수리영) 앞에 서서 무거운 눈빛으로 내려다보는 전신.\n\nLOCATION (lock): On the fishing boat's open deck beside the seated passenger, away from the clustered anglers. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the sunlit fishing-boat deck, sea backdrop, railings, and seated passenger area. Exclude the friendly fisherman from the immediate frame and place the stern man before the woman.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 5,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지정된 '전신' 프레이밍을 발끝까지 포함해 유일하게 충족했으며, '앞에 서서 내려다보는' 캐릭터 간의 배치와 시선을 가장 정확하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "남성의 발 부분이 화면에서 잘려 '전신' 숏 지시를 어겼으며, 여성의 앞이 아닌 옆에 서 있어 연출이 부정확합니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "'전신' 프레이밍 지시 위반과 잘못된 위치 배치 외에도, 레퍼런스의 올블랙 복장과 달리 남성의 이너 셔츠 색상이 다르게 렌더링되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S19sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지후의 무릎 위에 놓인 양손의 손가락 형태가 뭉개져 불분명합니다.",
     "fix_en": "Render clear, anatomically correct fingers for both of Jihu's hands resting on her lap.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "지후가 착용한 뿔뼈 펜던트의 크기가 비정상적으로 커서 나뭇가지처럼 보입니다.",
     "fix_en": "Scale down the antler-bone pendant to a realistic, wearable jewelry size while keeping its shape.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Render clear, anatomically correct fingers for both of Jihu's hands resting on her lap.\n- Scale down the antler-bone pendant to a realistic, wearable jewelry size while keeping its shape.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S19sh2"
  }
 },
 "S19sh12": {
  "input_fingerprint": "22f520aa7497ad7a",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 눈부신 햇살이 쏟아지는 푸른 물결 위로, 주변 섬들이 모두 사라진 채 홀로 신비롭게 떠 있는 금월도의 압도적인 전경.\n\nLOCATION (lock): Across the sunlit open sea ahead of the fishing boat, where a lone island rises beyond the glittering water. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 8,
   "C": 7,
   "A": 6
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "'눈부신 햇살이 쏟아지는'과 '신비롭게 떠 있는' 섬의 분위기를 가장 극적이고 충실하게 연출한 훌륭한 결과물입니다."
   },
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "지정된 섬의 형태와 반짝이는 바다를 잘 구현했으나, 빛이 쏟아지는 신비로운 분위기는 B에 비해 다소 약합니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "지정된 위치(어선 앞)를 배의 난간으로 표현한 점은 타당하나, 텍스트가 요구한 '홀로 떠 있는 압도적인 전경'에 집중하기에는 전경의 배가 다소 시선을 분산시킵니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L13B02.png"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S20sh5::bgfirst_bg": {
  "input_fingerprint": "46d8b160a419da97",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 과도한 환영 인파에 위축된 김지후(강수리영)가 어깨를 웅크린 채 당황한 눈빛으로 주변을 살피는 얼굴.\n\nLOCATION (lock): On the island pier immediately beside the newly arrived fishing boat and surrounding welcome crowd.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From shoulder height at the outer edge of the villagers, complete the dolly-in on a face-dominant frame of 김지후(강수리영) seen through restless bodies at both edges. Her hunched shoulders remain at the bottom of frame while her unsettled eyes search beyond the lens, and no other face intrudes into the central area.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 김지후(강수리영) isolated within the welcome crowd in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: pier walkway (Crowded by welcoming villagers) — The walkway recedes behind Jihoo toward the end of the pier; used as Supplies a narrow receding route behind Jihoo and situates the crowd's pressure; gathered crowd at frame edges (Clapping and cheering intensely); used as Appears only as fragmented peripheral motion that encloses Jihoo without displacing her face.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 과도한 환영 인파에 위축된 김지후(강수리영)가 어깨를 웅크린 채 당황한 눈빛으로 주변을 살피는 얼굴.\n\nLOCATION (lock): On the island pier immediately beside the newly arrived fishing boat and surrounding welcome crowd.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From shoulder height at the outer edge of the villagers, complete the dolly-in on a face-dominant frame of 김지후(강수리영) seen through restless bodies at both edges. Her hunched shoulders remain at the bottom of frame while her unsettled eyes search beyond the lens, and no other face intrudes into the central area.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 김지후(강수리영) isolated within the welcome crowd in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: pier walkway (Crowded by welcoming villagers) — The walkway recedes behind Jihoo toward the end of the pier; used as Supplies a narrow receding route behind Jihoo and situates the crowd's pressure; gathered crowd at frame edges (Clapping and cheering intensely); used as Appears only as fragmented peripheral motion that encloses Jihoo without displacing her face.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S20sh5__bgfirst_bg.png",
  "asset_id": "3962e3f1-5377-4ee8-94a0-8decec19e1ea",
  "input_asset_ids": [
   "ee58a65a-ed43-4b17-9699-d7ed92cb286b",
   "a553bc1d-51b8-4416-9cd3-f844bfb2f2b9"
  ]
 },
 "S20sh5": {
  "input_fingerprint": "6244e7ee006ddf41",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 과도한 환영 인파에 위축된 김지후(강수리영)가 어깨를 웅크린 채 당황한 눈빛으로 주변을 살피는 얼굴.\n\nLOCATION (lock): On the island pier immediately beside the newly arrived fishing boat and surrounding welcome crowd. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 과도한 환영 인파에 위축된 김지후(강수리영)가 어깨를 웅크린 채 당황한 눈빛으로 주변을 살피는 얼굴.\n\nLOCATION (lock): On the island pier immediately beside the newly arrived fishing boat and surrounding welcome crowd. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 과도한 환영 인파에 위축된 김지후(강수리영)가 어깨를 웅크린 채 당황한 눈빛으로 주변을 살피는 얼굴.\n\nLOCATION (lock): On the island pier immediately beside the newly arrived fishing boat and surrounding welcome crowd. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S20sh5__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S20sh5.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L15B02.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시문이 요구한 '얼굴' 중심의 클로즈업 프레이밍을 정확히 맞췄으며, 어깨를 웅크린 당황한 표정 연기도 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "'얼굴'을 묘사하라는 지시를 무시하고 전신이 드러나는 넓은 샷을 연출하여 프레이밍 조건(우선순위 2)을 크게 위반했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시문이 요구한 '얼굴' 중심의 클로즈업 프레이밍을 정확히 맞췄으며, 어깨를 웅크린 당황한 표정 연기도 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "'얼굴'을 묘사하라는 지시를 무시하고 전신이 드러나는 넓은 샷을 연출하여 프레이밍 조건(우선순위 2)을 크게 위반했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "프롬프트에 명시된 '얼굴' 중심의 구도와 어깨를 웅크리며 당황하는 표정 연기를 매우 정확하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "프롬프트가 요구한 '얼굴' 중심의 샷을 무시하고 전신이 보이는 와이드 샷으로 렌더링하여 프레이밍 우선순위를 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "프롬프트에 명시된 '얼굴' 중심의 구도와 어깨를 웅크리며 당황하는 표정 연기를 매우 정확하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "프롬프트가 요구한 '얼굴' 중심의 샷을 무시하고 전신이 보이는 와이드 샷으로 렌더링하여 프레이밍 우선순위를 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 7,
     "B": 15
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 7,
   "B": 15
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지시문이 요구한 '얼굴' 중심의 클로즈업 프레이밍을 정확히 맞췄으며, 어깨를 웅크린 당황한 표정 연기도 훌륭하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "'얼굴'을 묘사하라는 지시를 무시하고 전신이 드러나는 넓은 샷을 연출하여 프레이밍 조건(우선순위 2)을 크게 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L15B02.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "화면 왼쪽의 군중이 들고 있는 소형 깃발들에 파란색과 빨간색 텍스트가 포함되어 있어 '텍스트 없음(No text)' 지침을 위반했습니다.",
     "fix_en": "Remove all blue and red lettering from the small white flags held by the people on the left side of the frame, leaving the flags plain.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all blue and red lettering from the small white flags held by the people on the left side of the frame, leaving the flags plain.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S20sh5__bgfirst_bg.png",
   "bg_asset_id": "3962e3f1-5377-4ee8-94a0-8decec19e1ea",
   "bg_record_key": "S20sh5::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S20sh8": {
  "input_fingerprint": "7021177111b84270",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 60대 후반의 강영란(흰 머리 여자)이 온화한 미소를 지으며 선착장에 꼿꼿하게 서 있는 자태.\n\nLOCATION (lock): At the landward end of the island pier, facing the arriving passenger. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the colorful island harbor, gathered villagers, fishing boat, and bright daylight. Exclude the woman's overwhelmed close framing and isolate the elderly woman at the pier end.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 8,
   "A": 4,
   "B": 2
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "지정된 노파의 온화한 미소와 자세를 훌륭히 구현했으며, 이전 숏에 등장한 김지후의 뒷모습을 전경에 배치해 연속성을 완벽히 살렸습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "허용되지 않은 짧은 머리 남성의 뒷모습이 전경에 등장하여 등장인물 제한 지침을 위반했습니다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "화면 왼쪽 가장자리에 촬영용 카메라와 스태프가 그대로 노출되는 치명적인 오류가 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S20sh5_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1225321>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지문에서 노인을 단독으로 조명하라고(isolate the elderly woman) 지시했으나, 우측 전경에 이전 샷 여성의 뒷모습이 크게 배치되어 이를 위배합니다.",
     "fix_en": "Remove the dark-haired person in the right foreground entirely to isolate the elderly woman as the sole main subject.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "텍스트 생성 금지 지시에도 불구하고, 우측 전경의 파란 옷을 입은 주민이 든 태극기의 문양이 알파벳 'H'와 'I' 형태로 잘못 생성되었습니다.",
     "fix_en": "Redraw the markings on the flag held by the woman in blue on the right so they are plain black straight lines, removing the letter-like shapes.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the dark-haired person in the right foreground entirely to isolate the elderly woman as the sole main subject.\n- Redraw the markings on the flag held by the woman in blue on the right so they are plain black straight lines, removing the letter-like shapes.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S20sh5"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S20sh11": {
  "input_fingerprint": "51d38f3fc37a9493",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)이 김지후(강수리영)를 품에 밀착하여 꼭 끌어안고 있는 정지 상태.\n\nLOCATION (lock): On the crowded island pier, just beyond the fishing boat's disembarkation point. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the island pier, festive crowd, boat, and daylight established in the reference. Exclude the earlier separation between the two women and show their required embrace.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant while held in Youngran’s embrace.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 4,
   "A": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "프롬프트의 요구대로 두 인물의 외양과 의상을 올바르게 유지하며 포옹 장면을 사실적으로 구현했습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "두 인물의 의상(한복과 스웨터)이 서로 완전히 뒤바뀌어 캐릭터 정체성 유지에 실패했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "인물들의 의상이 뒤바뀌었으며, 스웨터 등 뒤에 목걸이가 붙어 있는 등 물리적 구조 오류가 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S20sh8_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "김지후의 가슴 쪽에 있어야 할 사슴 뿔/뼈 펜던트가 누락되고 목걸이 줄만 그려져 있습니다.",
     "fix_en": "Paint the large antler-bone pendant attached to the black cord on Jihu's chest, matching the character reference.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "김지후의 어깨를 감싼 강영란의 왼손이 해부학적으로 어색하게 그려졌습니다 (엄지손가락이 안쪽 목이 아닌 바깥쪽을 향하고 있어 오른손처럼 보임).",
     "fix_en": "Redraw the hand on Jihu's shoulder as a correct left hand, placing the thumb on the inside edge near her neck.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Paint the large antler-bone pendant attached to the black cord on Jihu's chest, matching the character reference.\n- Redraw the hand on Jihu's shoulder as a correct left hand, placing the thumb on the inside edge near her neck.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S20sh8"
  }
 },
 "groupbg::복합식당_홀": {
  "input_fingerprint": "c09620378e8c84cb",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "복합식당_홀",
    "tags": [
     "S21sh10",
     "S21sh16",
     "S21sh2",
     "S24sh4",
     "S24sh5",
     "S48sh10",
     "S48sh3",
     "S63sh14",
     "S63sh2",
     "S63sh8",
     "S83sh4",
     "S83sh5",
     "S90sh1",
     "S90sh4"
    ]
   },
   "context_sig": "1a95cc3f5de4bcf8"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: At a dining table inside a small multipurpose island restaurant and shop, lit by daytime through its windows.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n경찰 순찰정 외부 및 갑판: 바다를 달리는 흰색 바탕의 소형 경찰 순찰정. 선수와 선미를 잇는 좁은 통로가 있다. (특징: 하얀색과 파란색 도장의 선체; 갑판 바닥; 가장자리에 설치된 금속 레일 바; 바다의 하얀 물보라) / 금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인) / 금월도 해송 숲길 및 숲속: 해송이 울창하게 우거져 빛이 잘 들지 않는 빽빽하고 좁은 흙길과 숲의 심층부. (특징: 빽빽한 해송 나무 기둥들; 무성하고 어두운 수풀; 바닥에 깔린 흙과 솔잎; ‘수리영’이 적힌 이끼 낀 비석들; 비석 앞에 놓인 썩은 핏빛 육고기 덩어리) / 선착장 어구 창고 내부: 선착장 인근 낡은 창고 내부. 어망과 낚싯대, 생선 바구니가 천장까지 어지럽게 쌓여 비좁다. (특징: 천장까지 높게 쌓인 그물망과 통발 더미; 오래된 낚싯대들; 쌓여 있는 플라스틱 생선 바구니; 비좁고 어두운 먼지 낀 공간) / 해안마을 선착장: 육지 쪽 항구의 선착장. 어선들이 정박해 있고 콘크리트 바닥에 폴리스 라인이 쳐져 있다. (특징: 콘크리트 항구 부두 바닥; 드럼통 모닥불; 시체 주변의 노란색 폴리스 라인; 구경꾼 무리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실내. 반월 식당 - 낮\n- 말이 식당이지 초등학교 교실 정도 크기에 식당과 편의점, 낚시 용품점 등을 총 망라한 잡화점.\n- 실내. 반월 식당 - 밤\n- 실내. 반월 식당 - 낮\n- 실내. 반월 식당 - 밤\n- 실내. 반월 식당 - 낮\n- 실내. 반월 식당 - 해질녘\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: At a dining table inside a small multipurpose island restaurant and shop, lit by daytime through its windows.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n경찰 순찰정 외부 및 갑판: 바다를 달리는 흰색 바탕의 소형 경찰 순찰정. 선수와 선미를 잇는 좁은 통로가 있다. (특징: 하얀색과 파란색 도장의 선체; 갑판 바닥; 가장자리에 설치된 금속 레일 바; 바다의 하얀 물보라) / 금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인) / 금월도 해송 숲길 및 숲속: 해송이 울창하게 우거져 빛이 잘 들지 않는 빽빽하고 좁은 흙길과 숲의 심층부. (특징: 빽빽한 해송 나무 기둥들; 무성하고 어두운 수풀; 바닥에 깔린 흙과 솔잎; ‘수리영’이 적힌 이끼 낀 비석들; 비석 앞에 놓인 썩은 핏빛 육고기 덩어리) / 선착장 어구 창고 내부: 선착장 인근 낡은 창고 내부. 어망과 낚싯대, 생선 바구니가 천장까지 어지럽게 쌓여 비좁다. (특징: 천장까지 높게 쌓인 그물망과 통발 더미; 오래된 낚싯대들; 쌓여 있는 플라스틱 생선 바구니; 비좁고 어두운 먼지 낀 공간) / 해안마을 선착장: 육지 쪽 항구의 선착장. 어선들이 정박해 있고 콘크리트 바닥에 폴리스 라인이 쳐져 있다. (특징: 콘크리트 항구 부두 바닥; 드럼통 모닥불; 시체 주변의 노란색 폴리스 라인; 구경꾼 무리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실내. 반월 식당 - 낮\n- 말이 식당이지 초등학교 교실 정도 크기에 식당과 편의점, 낚시 용품점 등을 총 망라한 잡화점.\n- 실내. 반월 식당 - 밤\n- 실내. 반월 식당 - 낮\n- 실내. 반월 식당 - 밤\n- 실내. 반월 식당 - 낮\n- 실내. 반월 식당 - 해질녘\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_복합식당_홀_7d32c1.png",
  "asset_id": "9dda17be-ce32-45ab-a790-af311277afb2",
  "input_asset_ids": [
   "b490755f-a3ac-4986-ab07-611b467cb6ac"
  ],
  "origin_tag": "S21sh2",
  "place_text": "At a dining table inside a small multipurpose island restaurant and shop, lit by daytime through its windows.",
  "origin_inputs": {
   "place_text": "At a dining table inside a small multipurpose island restaurant and shop, lit by daytime through its windows.",
   "time_of_day_en": "day",
   "conti_asset_id": "b490755f-a3ac-4986-ab07-611b467cb6ac"
  }
 },
 "S21sh2::bgfirst_bg": {
  "input_fingerprint": "c72214cea6378c5f",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 식당 테이블에 나란히 앉은 강영란(흰 머리 여자)이 김지후(강수리영) 곁에서 낡은 사진첩을 짚어 보이는 모습.\n\nLOCATION (lock): At a dining table inside a small multipurpose island restaurant and shop, lit by daytime through its windows.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At seated eye level beside the table, continue the gentle dolly-in but stop slightly wider than the stage endpoint so 강영란(흰 머리 여자), 김지후(강수리영), and the open album remain readable together. Youngran presses close on the left and points into the album at lower center, while Jihoo leans in from the right with the mixed restaurant-store interior held in soft depth behind them.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: open old photo album (Open on the table as Youngran points through it) — The open image-bearing pages face upward and partly toward the camera, displaying photographs of young Minsook and Jihoo as a baby; used as Shared focal evidence positioned below both women's faces; dining table (Supporting the open album) — The near edge crosses the lower frame and recedes toward the kitchen side; used as Anchors the two seated figures and guides the later arc toward the kitchen; combined restaurant and general-store interior (Includes dining, convenience goods, and fishing supplies); used as Provides the broader combined function of the location behind the intimate album viewing.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 식당 테이블에 나란히 앉은 강영란(흰 머리 여자)이 김지후(강수리영) 곁에서 낡은 사진첩을 짚어 보이는 모습.\n\nLOCATION (lock): At a dining table inside a small multipurpose island restaurant and shop, lit by daytime through its windows.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At seated eye level beside the table, continue the gentle dolly-in but stop slightly wider than the stage endpoint so 강영란(흰 머리 여자), 김지후(강수리영), and the open album remain readable together. Youngran presses close on the left and points into the album at lower center, while Jihoo leans in from the right with the mixed restaurant-store interior held in soft depth behind them.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: open old photo album (Open on the table as Youngran points through it) — The open image-bearing pages face upward and partly toward the camera, displaying photographs of young Minsook and Jihoo as a baby; used as Shared focal evidence positioned below both women's faces; dining table (Supporting the open album) — The near edge crosses the lower frame and recedes toward the kitchen side; used as Anchors the two seated figures and guides the later arc toward the kitchen; combined restaurant and general-store interior (Includes dining, convenience goods, and fishing supplies); used as Provides the broader combined function of the location behind the intimate album viewing.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S21sh2__bgfirst_bg.png",
  "asset_id": "a08dfb2c-b932-433f-b04d-7d3450e7278b",
  "input_asset_ids": [
   "b490755f-a3ac-4986-ab07-611b467cb6ac",
   "9dda17be-ce32-45ab-a790-af311277afb2"
  ]
 },
 "S21sh2": {
  "input_fingerprint": "b91fe07733b96da2",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 식당 테이블에 나란히 앉은 강영란(흰 머리 여자)이 김지후(강수리영) 곁에서 낡은 사진첩을 짚어 보이는 모습.\n\nLOCATION (lock): At a dining table inside a small multipurpose island restaurant and shop, lit by daytime through its windows. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비); 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 식당 테이블에 나란히 앉은 강영란(흰 머리 여자)이 김지후(강수리영) 곁에서 낡은 사진첩을 짚어 보이는 모습.\n\nLOCATION (lock): At a dining table inside a small multipurpose island restaurant and shop, lit by daytime through its windows. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비); 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 식당 테이블에 나란히 앉은 강영란(흰 머리 여자)이 김지후(강수리영) 곁에서 낡은 사진첩을 짚어 보이는 모습.\n\nLOCATION (lock): At a dining table inside a small multipurpose island restaurant and shop, lit by daytime through its windows. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비); 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S21sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S21sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:867030>"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_복합식당_홀_7d32c1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:867030>"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "두 인물의 외모와 의상이 레퍼런스와 정확히 일치하며, 사진첩 내부의 사진들이 캐릭터들을 향해 올바르게 배열되어 프롬프트를 완벽히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "사진첩 내의 사진들이 카메라를 향하거나 옆으로 누워 있어 소품 방향 규칙을 위반했으며, 강영란의 의상이 레퍼런스와 일치하지 않습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "두 인물의 외모와 의상이 레퍼런스와 정확히 일치하며, 사진첩 내부의 사진들이 캐릭터들을 향해 올바르게 배열되어 프롬프트를 완벽히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "사진첩 내의 사진들이 카메라를 향하거나 옆으로 누워 있어 소품 방향 규칙을 위반했으며, 강영란의 의상이 레퍼런스와 일치하지 않습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지정된 인물들의 외모, 의상 및 펜던트 소품을 레퍼런스와 정확히 일치시켰으며, 사진첩을 짚어 보이는 동작과 배경을 충실히 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "배경 주방 쪽에 프롬프트에 없는 인물이 등장하여 심각한 오류(Hard Violation)이며, 두 인물의 의상과 펜던트 디자인이 레퍼런스와 다릅니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 인물들의 외모, 의상 및 펜던트 소품을 레퍼런스와 정확히 일치시켰으며, 사진첩을 짚어 보이는 동작과 배경을 충실히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "배경 주방 쪽에 프롬프트에 없는 인물이 등장하여 심각한 오류(Hard Violation)이며, 두 인물의 의상과 펜던트 디자인이 레퍼런스와 다릅니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "두 인물의 외모와 의상이 레퍼런스와 정확히 일치하며, 사진첩 내부의 사진들이 캐릭터들을 향해 올바르게 배열되어 프롬프트를 완벽히 구현했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "사진첩 내의 사진들이 카메라를 향하거나 옆으로 누워 있어 소품 방향 규칙을 위반했으며, 강영란의 의상이 레퍼런스와 일치하지 않습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_복합식당_홀_7d32c1.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:867030>"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "사진첩 안의 사진들이 카메라 방향으로 똑바로 놓여 있어, 책을 보고 있는 등장인물들에게는 위아래가 뒤집혀 있습니다.",
     "fix_en": "Rotate the photographs inside the album 180 degrees so they are oriented right-side up for the characters looking at them.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "사진첩의 왼쪽 페이지 여백에 손글씨 텍스트가 적혀 있습니다.",
     "fix_en": "Remove the handwritten text from the photo album page, replacing it with plain paper texture.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Rotate the photographs inside the album 180 degrees so they are oriented right-side up for the characters looking at them.\n- Remove the handwritten text from the photo album page, replacing it with plain paper texture.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S21sh2__bgfirst_bg.png",
   "bg_asset_id": "a08dfb2c-b932-433f-b04d-7d3450e7278b",
   "bg_record_key": "S21sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "복합식당_홀",
   "groupbg_asset_id": "9dda17be-ce32-45ab-a790-af311277afb2"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S21sh10": {
  "input_fingerprint": "602ffe4c87194221",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 고산들이 선홍빛 핏기가 선명한 육회와 생고기가 잔뜩 담긴 쟁반을 테이블 위로 반쯤 내려놓은 채 정지한 자세.\n\nLOCATION (lock): Beside the dining table in the restaurant's main room, between the kitchen and seated guests. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the modest multipurpose restaurant interior, table, windows, and daylight. Exclude the open photo album action and add only the required meat dishes and server.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears the finger-length, antler-bone pendant at her neck.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 5,
   "A": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "명시된 인물만 화면에 등장시키고 이전 샷의 빈 배경을 그대로 유지하여 프롬프트를 가장 정확하게 구현함."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "주요 인물의 행동과 복장은 적절하나, 지시문에 없는 노인과 배경 인물들을 임의로 추가하여 감점됨."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "캐릭터의 복장이 참조 이미지와 다르고, 배경에 텍스트가 요구하지 않은 인물들을 생성하여 규칙을 위반함."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S21sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "쟁반이 테이블 표면에 완전히 닿아 있습니다. 텍스트에는 '반쯤 내려놓은 채'로 공중에 떠 있어야 한다고 명시되어 있습니다.",
     "fix_en": "Elevate the tray and the hands holding it so that it hovers slightly above the table surface, with a gap and shadow indicating it is halfway lowered.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "화면 오른쪽, 쟁반을 잡고 있는 남성의 왼손 손가락 구조가 무너져 쟁반 가장자리와 하나로 뭉개져 있습니다.",
     "fix_en": "Redraw the man's left hand gripping the right side of the tray to have anatomically correct, distinct fingers resting clearly on the edge.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Elevate the tray and the hands holding it so that it hovers slightly above the table surface, with a gap and shadow indicating it is halfway lowered.\n- Redraw the man's left hand gripping the right side of the tray to have anatomically correct, distinct fingers resting clearly on the edge.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S21sh2"
  }
 },
 "S21sh16": {
  "input_fingerprint": "01a00e7a63668aee",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 붉은 피를 입가에 잔뜩 묻힌 채 짐승처럼 고기를 뜯어내기 위해 고개를 뒤로 젖힌 자세의 금월도 꼬마 아이들 클로즈업.\n\nLOCATION (lock): Around the food-laden table inside the island restaurant's dining room, in daylight from the windows. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the restaurant table area, daylight, and abundant raw-meat spread. Exclude the two adult women from the close frame and show only the children eating ravenously.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The children’s mouths and hands remain smeared with blood from eating the raw meat; Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 6,
   "B": 4,
   "A": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "유일하게 '고개를 뒤로 젖힌' 동작을 부분적으로 구현했으나, 생고기 대신 익힌 고기가 등장하고 소품(목걸이)이 중복되었습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "아이들만 프레임에 담았으나 고개를 젖힌 핵심 동작이 없으며, 생고기 지시를 어기고 꼬치구이가 등장했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "아이들만 보여달라는 지시를 어기고 어른이 포함되었으며(좌측), 요구된 동작과 생고기 설정 모두 누락되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S21sh10_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:902182>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 컷의 연속성 및 프롬프트에 명시된 붉은 생고기 대신 조리된 갈색 고기와 밥그릇, 반찬 등이 식탁에 놓여 있습니다.",
     "fix_en": "Replace the platter of cooked brown meat, rice bowls, soup bowls, and side dishes on the table with plates filled entirely with bright red raw meat.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "아이들이 짐승처럼 고기를 뜯어먹어야 함에도 젓가락을 사용하고 있으며, 손에 피가 묻어있지 않습니다.",
     "fix_en": "Remove all chopsticks and redraw the children's hands to be directly grasping pieces of raw meat, adding heavy smears of bright red blood to their hands and mouths.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "모든 아이들의 고개가 뒤로 젖혀진 자세여야 하나, 가운데 소년과 오른쪽 소녀는 정면이나 아래를 향하고 있습니다.",
     "fix_en": "Tilt the heads of the center boy and the right-side girl backward so their faces point upward, matching the pose of the girl on the left.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "단일 인물(지후)의 특징인 뼈 목걸이를 왼쪽과 오른쪽 두 소녀가 동일하게 착용하고 있습니다.",
     "fix_en": "Erase the antler-bone pendant from the chest of the girl on the right, leaving only her plain shirt.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the platter of cooked brown meat, rice bowls, soup bowls, and side dishes on the table with plates filled entirely with bright red raw meat.\n- Remove all chopsticks and redraw the children's hands to be directly grasping pieces of raw meat, adding heavy smears of bright red blood to their hands and mouths.\n- Erase the antler-bone pendant from the chest of the girl on the right, leaving only her plain shirt.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S21sh10"
  }
 },
 "S22sh2": {
  "input_fingerprint": "a114bc0f3d414b3b",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): sunset transitioning to night.\n\nSHOT TEXT (authoritative, Korean): 짙은 어둠 속, 마을 곳곳에 피어 있던 꽃들이 바짝 말라 비틀어진 기이한 밤의 상태.\n\nLOCATION (lock): Across the flowered outdoor corners of the small island village after nightfall. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The flowers across the village have withered and remain shriveled after nightfall.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 8,
   "B": 6,
   "A": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "레퍼런스의 공간 구조(길목과 화단)를 그대로 유지하면서 바짝 마른 꽃들의 기이한 상태를 잘 표현했습니다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "꽃이 마른 상태는 잘 구현되었으나, 우측 길 한가운데에 레퍼런스에 없는 거대한 화분을 추가하여 공간을 변형시켰습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "길목에 존재하지 않는 큰 화분을 임의로 추가했으며, 꽃이 마른 것을 넘어 까맣게 타버린 듯한 비현실적인 형태로 묘사되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L18B04.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "왼쪽 벽의 파란색 표지판에 텍스트와 숫자 '3'이 있습니다 (프롬프트에서 모든 텍스트 포함 금지).",
     "fix_en": "Remove the white text and the number '3' from the blue plaque on the left wall, leaving it completely blank."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the white text and the number '3' from the blue plaque on the left wall, leaving it completely blank.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "groupbg::해송숲_오솔길": {
  "input_fingerprint": "5b0bac8546656043",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "해송숲_오솔길",
    "tags": [
     "S107sh3",
     "S23sh1",
     "S29sh1",
     "S29sh3",
     "S29sh5",
     "S33sh12",
     "S33sh13",
     "S33sh6",
     "S40sh1",
     "S40sh8",
     "S45sh10",
     "S45sh3",
     "S45sh8",
     "S64sh2",
     "S67sh3",
     "S67sh8",
     "S67sh9",
     "S69sh10",
     "S69sh4",
     "S69sh9",
     "S74sh2",
     "S74sh8"
    ]
   },
   "context_sig": "18b3e62c1cbdb7ef"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: Along a narrow path beneath dense pine trees, lit only by faint moonlight.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 숲속 만월제 도당: 숲 한가운데 둥글게 트인 공터. 거대한 바위와 제단이 있고 커다란 모닥불들이 타오른다. (특징: 성인 키 서너 배 높이의 거대한 암석 바위; 피 손자국으로 범벅이 된 바위 표면; 바위 앞 제단; 제단을 둥글게 감싸며 타오르는 커다란 모닥불들; 하늘 위 거대한 핏빛 보름달) / 금월도 해안 마을: 50여 가구가 채 안 되는 섬마을. 비탈을 따라 알록달록한 지붕의 소담스러운 집들이 단풍 든 나무와 어우러져 있다. (특징: 좁고 야트막한 담장의 마을 골목길; 알록달록한 지붕의 낡은 집들; 집 앞 대문에 걸린 새끼줄과 종이 인형; 마당과 담장에 심어진 짙은 녹색 석창포 약초) / 반월 식당 내부: 초등학교 교실 크기에 식당, 편의점, 낚시 용품점이 섞인 잡화점 느낌의 식당. 나무 테이블과 주방, 진열대가 혼재되어 있다. (특징: 다수의 낡은 목재 테이블; 격자무늬 나무 유리창; 개방된 주방 공간; 낚시 용품과 잡화 진열대; 붉은 피가 흐르는 뭉티기와 생간 등 날고기 한 상) / 반월 식당 뒤뜰: 식당 건물 뒤편의 흙마당. 잘 꾸며진 텃밭과 평상이 있고 낡은 창고 건물이 보인다. (특징: 흙바닥 텃밭; 나무 평상; 말려둔 시래기 묶음; 허름한 창고 건물의 목재 외벽) / 반월 식당 앞: 식당 건물 앞의 골목길 및 입구 흙길 지형. (특징: 미닫이 나무 문; 낡은 식당 외벽; 흙 덮인 골목길; 바닥의 짙은 핏자국) / 숲속 도축장 내부: 벽에 도축용 기구들이 널려 있는 가건물 내부. 바닥과 벽면에 검붉은 핏자국이 선명하다. (특징: 도축용 칼과 뾰족한 금속 갈고리들; 피받이 양동이; 핏자국이 흥건한 도축대, 바닥, 벽면; 구석에 쌓인 피 묻은 포대자루 더미; 포대자루에서 쏟아진 토막 난 인체 부위) / 숲속 도축장 외부 및 진입로: 숲속 깊은 곳 염소 축사 옆에 위치한 나무 가건물 외관. (특징: 얼기설기 대충 지은 나무 가건물 외벽; 흙 묻고 낡은 판자벽; 유리가 깨져 먼지와 거미줄이 낀 창문)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 숲길 - 밤\n- 어슴프레 달빛만이 비치는 울창한 해송 나무 숲\n- 실외. 숲길 - 밤\n- 실외. 숲속 입구 - 낮\n- 실외. 금월도 숲속 - 밤\n- 실외. 숲길 - 밤\n- 다시 숲속 - 밤\n- 실외. 숲속 - 밤\n- 실외. 숲 속 - 밤\n\nTIME OF DAY (lock): night, faint moonlight.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: Along a narrow path beneath dense pine trees, lit only by faint moonlight.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 숲속 만월제 도당: 숲 한가운데 둥글게 트인 공터. 거대한 바위와 제단이 있고 커다란 모닥불들이 타오른다. (특징: 성인 키 서너 배 높이의 거대한 암석 바위; 피 손자국으로 범벅이 된 바위 표면; 바위 앞 제단; 제단을 둥글게 감싸며 타오르는 커다란 모닥불들; 하늘 위 거대한 핏빛 보름달) / 금월도 해안 마을: 50여 가구가 채 안 되는 섬마을. 비탈을 따라 알록달록한 지붕의 소담스러운 집들이 단풍 든 나무와 어우러져 있다. (특징: 좁고 야트막한 담장의 마을 골목길; 알록달록한 지붕의 낡은 집들; 집 앞 대문에 걸린 새끼줄과 종이 인형; 마당과 담장에 심어진 짙은 녹색 석창포 약초) / 반월 식당 내부: 초등학교 교실 크기에 식당, 편의점, 낚시 용품점이 섞인 잡화점 느낌의 식당. 나무 테이블과 주방, 진열대가 혼재되어 있다. (특징: 다수의 낡은 목재 테이블; 격자무늬 나무 유리창; 개방된 주방 공간; 낚시 용품과 잡화 진열대; 붉은 피가 흐르는 뭉티기와 생간 등 날고기 한 상) / 반월 식당 뒤뜰: 식당 건물 뒤편의 흙마당. 잘 꾸며진 텃밭과 평상이 있고 낡은 창고 건물이 보인다. (특징: 흙바닥 텃밭; 나무 평상; 말려둔 시래기 묶음; 허름한 창고 건물의 목재 외벽) / 반월 식당 앞: 식당 건물 앞의 골목길 및 입구 흙길 지형. (특징: 미닫이 나무 문; 낡은 식당 외벽; 흙 덮인 골목길; 바닥의 짙은 핏자국) / 숲속 도축장 내부: 벽에 도축용 기구들이 널려 있는 가건물 내부. 바닥과 벽면에 검붉은 핏자국이 선명하다. (특징: 도축용 칼과 뾰족한 금속 갈고리들; 피받이 양동이; 핏자국이 흥건한 도축대, 바닥, 벽면; 구석에 쌓인 피 묻은 포대자루 더미; 포대자루에서 쏟아진 토막 난 인체 부위) / 숲속 도축장 외부 및 진입로: 숲속 깊은 곳 염소 축사 옆에 위치한 나무 가건물 외관. (특징: 얼기설기 대충 지은 나무 가건물 외벽; 흙 묻고 낡은 판자벽; 유리가 깨져 먼지와 거미줄이 낀 창문)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 숲길 - 밤\n- 어슴프레 달빛만이 비치는 울창한 해송 나무 숲\n- 실외. 숲길 - 밤\n- 실외. 숲속 입구 - 낮\n- 실외. 금월도 숲속 - 밤\n- 실외. 숲길 - 밤\n- 다시 숲속 - 밤\n- 실외. 숲속 - 밤\n- 실외. 숲 속 - 밤\n\nTIME OF DAY (lock): night, faint moonlight.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송숲_오솔길_22bdfa.png",
  "asset_id": "09cf0b8b-8857-49e2-af2b-fcea7222c63c",
  "input_asset_ids": [
   "a987e339-1724-4e6f-a9ad-ababe0201bcf"
  ],
  "origin_tag": "S23sh1",
  "place_text": "Along a narrow path beneath dense pine trees, lit only by faint moonlight.",
  "origin_inputs": {
   "place_text": "Along a narrow path beneath dense pine trees, lit only by faint moonlight.",
   "time_of_day_en": "night, faint moonlight",
   "conti_asset_id": "a987e339-1724-4e6f-a9ad-ababe0201bcf"
  }
 },
 "S23sh1::bgfirst_bg": {
  "input_fingerprint": "c78ab5cdeaf9697d",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어스름한 달빛이 스며드는 빽빽한 해송 숲길, 한 발을 허공에 띄운 채 다음 걸음을 내디디려는 mid-stride 자세로 포착된 고산들의 전신.\n\nLOCATION (lock): Along a narrow path beneath dense pine trees, lit only by faint moonlight.\n\nTIME OF DAY (lock): night, faint moonlight.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track backward at approximately waist height, slightly offset from the path center, keeping 고산들 full-length as he advances rapidly toward the camera. Time the wide frame to the suspended beat between steps, with one foot lifted and his body pitched forward while the dense pine path recedes behind him.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 고산들 in the middle-center of the frame, midground, moves toward forward along the forest path.\n- KEY BACKGROUND ELEMENTS: 빽빽한 해송 숲 (울창함) — The trunks flank both sides of the frame and open into the path behind 고산들; used as Forms a dense corridor around the traveler and emphasizes his uninterrupted forward pace; 숲길 (불빛 하나 없이 어두움); used as Defines the travel axis along which the camera and character move in opposite directions.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어스름한 달빛이 스며드는 빽빽한 해송 숲길, 한 발을 허공에 띄운 채 다음 걸음을 내디디려는 mid-stride 자세로 포착된 고산들의 전신.\n\nLOCATION (lock): Along a narrow path beneath dense pine trees, lit only by faint moonlight.\n\nTIME OF DAY (lock): night, faint moonlight.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track backward at approximately waist height, slightly offset from the path center, keeping 고산들 full-length as he advances rapidly toward the camera. Time the wide frame to the suspended beat between steps, with one foot lifted and his body pitched forward while the dense pine path recedes behind him.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 고산들 in the middle-center of the frame, midground, moves toward forward along the forest path.\n- KEY BACKGROUND ELEMENTS: 빽빽한 해송 숲 (울창함) — The trunks flank both sides of the frame and open into the path behind 고산들; used as Forms a dense corridor around the traveler and emphasizes his uninterrupted forward pace; 숲길 (불빛 하나 없이 어두움); used as Defines the travel axis along which the camera and character move in opposite directions.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S23sh1__bgfirst_bg.png",
  "asset_id": "023f2cc6-a454-4ebd-a258-00d1034e5b06",
  "input_asset_ids": [
   "a987e339-1724-4e6f-a9ad-ababe0201bcf",
   "09cf0b8b-8857-49e2-af2b-fcea7222c63c"
  ]
 },
 "S23sh1": {
  "input_fingerprint": "5d72d5e69df97be0",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, faint moonlight.\n\nSHOT TEXT (authoritative, Korean): 어스름한 달빛이 스며드는 빽빽한 해송 숲길, 한 발을 허공에 띄운 채 다음 걸음을 내디디려는 mid-stride 자세로 포착된 고산들의 전신.\n\nLOCATION (lock): Along a narrow path beneath dense pine trees, lit only by faint moonlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, faint moonlight.\n\nSHOT TEXT (authoritative, Korean): 어스름한 달빛이 스며드는 빽빽한 해송 숲길, 한 발을 허공에 띄운 채 다음 걸음을 내디디려는 mid-stride 자세로 포착된 고산들의 전신.\n\nLOCATION (lock): Along a narrow path beneath dense pine trees, lit only by faint moonlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, faint moonlight.\n\nSHOT TEXT (authoritative, Korean): 어스름한 달빛이 스며드는 빽빽한 해송 숲길, 한 발을 허공에 띄운 채 다음 걸음을 내디디려는 mid-stride 자세로 포착된 고산들의 전신.\n\nLOCATION (lock): Along a narrow path beneath dense pine trees, lit only by faint moonlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S23sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S23sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송숲_오솔길_22bdfa.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시된 숲길을 따라 걷는 방향이 자연스러우며, 한 발을 내디디려는 전신 포즈와 역사적 맥락(1670년 조선, 고산)을 충실히 구현했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "공중에 뜬 발의 동작은 명확하나 숲길을 가로질러 걷는 구도가 장소의 방향성과 어긋나며, 캐릭터의 시대적 배경 설정이 반영되지 않았습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시된 숲길을 따라 걷는 방향이 자연스러우며, 한 발을 내디디려는 전신 포즈와 역사적 맥락(1670년 조선, 고산)을 충실히 구현했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "공중에 뜬 발의 동작은 명확하나 숲길을 가로질러 걷는 구도가 장소의 방향성과 어긋나며, 캐릭터의 시대적 배경 설정이 반영되지 않았습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 숲길 배경에서 한 발을 허공에 띄운 mid-stride 전신 샷을 정확히 연출했으며, 1670년 조선 시대라는 전체 맥락에 완벽히 부합합니다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "배경과 인물의 포즈 등 샷의 물리적 요구사항은 잘 구현했으나, 조선 시대 회상이라는 시대적 맥락이 누락된 현대 복장이라 아쉽습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 숲길 배경에서 한 발을 허공에 띄운 mid-stride 전신 샷을 정확히 연출했으며, 1670년 조선 시대라는 전체 맥락에 완벽히 부합합니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "배경과 인물의 포즈 등 샷의 물리적 요구사항은 잘 구현했으나, 조선 시대 회상이라는 시대적 맥락이 누락된 현대 복장이라 아쉽습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 10
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 10
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지시된 숲길을 따라 걷는 방향이 자연스러우며, 한 발을 내디디려는 전신 포즈와 역사적 맥락(1670년 조선, 고산)을 충실히 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "공중에 뜬 발의 동작은 명확하나 숲길을 가로질러 걷는 구도가 장소의 방향성과 어긋나며, 캐릭터의 시대적 배경 설정이 반영되지 않았습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송숲_오솔길_22bdfa.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "레이아웃 스케치에는 오른팔을 뒤로 뻗고 왼쪽 뒷발을 높이 든 역동적인 달리기 자세가 묘사되어 있으나, 생성된 이미지에서는 팔이 아래로 향하고 발이 거의 들리지 않은 밋밋한 걷기 자세를 취하고 있습니다.",
     "fix_en": "Swing the character's right arm backward and bend the left knee to raise the back foot higher, matching the dynamic running pose of the layout sketch.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Swing the character's right arm backward and bend the left knee to raise the back foot higher, matching the dynamic running pose of the layout sketch.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S23sh1__bgfirst_bg.png",
   "bg_asset_id": "023f2cc6-a454-4ebd-a258-00d1034e5b06",
   "bg_record_key": "S23sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "해송숲_오솔길",
   "groupbg_asset_id": "09cf0b8b-8857-49e2-af2b-fcea7222c63c"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S24sh4::bgfirst_bg": {
  "input_fingerprint": "9a8d9b27bdfef147",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 굳게 닫힌 채 불이 꺼져 있는 강영란(흰 머리 여자)의 방문을 쳐다보는 김지후(강수리영)의 뒷모습.\n\nLOCATION (lock): Inside the dark restaurant hall, facing the closed, unlit door of the adjoining bedroom; only weak moonlight enters through the windows.\n\nTIME OF DAY (lock): night, faint moonlight.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at shoulder height behind 김지후(강수리영), framing her at medium distance with her back in the lower center and 강영란(흰 머리 여자)의 closed room door directly beyond. Her raised shoulders and slight forward lean create a taut line of attention toward the unresponsive door, while enough of the dark dining room remains at the sides to register the absence of others.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 김지후(강수리영) in the lower-center of the frame, midground, looks toward closed room door; closed room door in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 강영란(흰 머리 여자)의 방문 (굳게 닫혀 있고 방의 불이 꺼져 있음) — Its closed exterior face is presented directly beyond the character; used as The fixed visual destination beyond 김지후(강수리영)'s back; 식당 창 (창으로 달빛이 스며들고 있음) — Moonlight passes through it into the otherwise dark restaurant; used as Provides the dim spatial context around the character and door.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 굳게 닫힌 채 불이 꺼져 있는 강영란(흰 머리 여자)의 방문을 쳐다보는 김지후(강수리영)의 뒷모습.\n\nLOCATION (lock): Inside the dark restaurant hall, facing the closed, unlit door of the adjoining bedroom; only weak moonlight enters through the windows.\n\nTIME OF DAY (lock): night, faint moonlight.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at shoulder height behind 김지후(강수리영), framing her at medium distance with her back in the lower center and 강영란(흰 머리 여자)의 closed room door directly beyond. Her raised shoulders and slight forward lean create a taut line of attention toward the unresponsive door, while enough of the dark dining room remains at the sides to register the absence of others.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 김지후(강수리영) in the lower-center of the frame, midground, looks toward closed room door; closed room door in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 강영란(흰 머리 여자)의 방문 (굳게 닫혀 있고 방의 불이 꺼져 있음) — Its closed exterior face is presented directly beyond the character; used as The fixed visual destination beyond 김지후(강수리영)'s back; 식당 창 (창으로 달빛이 스며들고 있음) — Moonlight passes through it into the otherwise dark restaurant; used as Provides the dim spatial context around the character and door.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S24sh4__bgfirst_bg.png",
  "asset_id": "0fdcd1d3-cee7-4a45-a6ec-9e672086b152",
  "input_asset_ids": [
   "6252f5e3-8581-4887-a6fe-92498ea9b4ca",
   "803cc956-9eff-49c4-b977-3e62ad55fc55"
  ]
 },
 "S24sh4": {
  "input_fingerprint": "e3f298fc1c82b0ed",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, faint moonlight.\n\nSHOT TEXT (authoritative, Korean): 굳게 닫힌 채 불이 꺼져 있는 강영란(흰 머리 여자)의 방문을 쳐다보는 김지후(강수리영)의 뒷모습.\n\nLOCATION (lock): Inside the dark restaurant hall, facing the closed, unlit door of the adjoining bedroom; only weak moonlight enters through the windows. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, faint moonlight.\n\nSHOT TEXT (authoritative, Korean): 굳게 닫힌 채 불이 꺼져 있는 강영란(흰 머리 여자)의 방문을 쳐다보는 김지후(강수리영)의 뒷모습.\n\nLOCATION (lock): Inside the dark restaurant hall, facing the closed, unlit door of the adjoining bedroom; only weak moonlight enters through the windows. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, faint moonlight.\n\nSHOT TEXT (authoritative, Korean): 굳게 닫힌 채 불이 꺼져 있는 강영란(흰 머리 여자)의 방문을 쳐다보는 김지후(강수리영)의 뒷모습.\n\nLOCATION (lock): Inside the dark restaurant hall, facing the closed, unlit door of the adjoining bedroom; only weak moonlight enters through the windows. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S24sh4__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S24sh4.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "식당 내부와 인물의 뒷모습을 지시대로 잘 구현했으나, 펜던트를 등 쪽으로 착용한 모습이 물리적으로 다소 어색합니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "실내 공간인 식당 내부에 야외 숲 레퍼런스 이미지가 그대로 합성된 물리적으로 불가능한 콜라주가 발생하여 심각한 규정 위반입니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "식당 내부와 인물의 뒷모습을 지시대로 잘 구현했으나, 펜던트를 등 쪽으로 착용한 모습이 물리적으로 다소 어색합니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "실내 공간인 식당 내부에 야외 숲 레퍼런스 이미지가 그대로 합성된 물리적으로 불가능한 콜라주가 발생하여 심각한 규정 위반입니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "식당 내부 공간에 야외 숲과 비석 이미지가 콜라주처럼 합성되어 있어 물리적으로 불가능한 공간(우선순위 1 위반)입니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "어두운 실내에서 닫힌 방문을 바라보는 인물의 뒷모습과 창문을 통한 달빛 등 지시문을 사실적이고 충실하게 구현했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "식당 내부 공간에 야외 숲과 비석 이미지가 콜라주처럼 합성되어 있어 물리적으로 불가능한 공간(우선순위 1 위반)입니다."
     },
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "어두운 실내에서 닫힌 방문을 바라보는 인물의 뒷모습과 창문을 통한 달빛 등 지시문을 사실적이고 충실하게 구현했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 6
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "식당 내부와 인물의 뒷모습을 지시대로 잘 구현했으나, 펜던트를 등 쪽으로 착용한 모습이 물리적으로 다소 어색합니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "실내 공간인 식당 내부에 야외 숲 레퍼런스 이미지가 그대로 합성된 물리적으로 불가능한 콜라주가 발생하여 심각한 규정 위반입니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "목걸이의 펜던트가 가슴 쪽이 아닌 등 뒤로 잘못 매달려 있습니다.",
     "fix_en": "Remove the antler-bone pendant from her back, leaving only the necklace cord wrapping around the back of her neck.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the antler-bone pendant from her back, leaving only the necklace cord wrapping around the back of her neck.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S24sh4__bgfirst_bg.png",
   "bg_asset_id": "0fdcd1d3-cee7-4a45-a6ec-9e672086b152",
   "bg_record_key": "S24sh4::bgfirst_bg",
   "chain_winner": true,
   "authority": "seed_bg"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S24sh5": {
  "input_fingerprint": "281afe8ebfdd4fcf",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, faint moonlight.\n\nSHOT TEXT (authoritative, Korean): 아무도 없는 텅 빈 식당 한가운데서 불안감에 휩싸인 채 어깨를 한껏 치켜올리며 숨을 참고 있는 김지후(강수리영)의 전신.\n\nLOCATION (lock): In the empty central dining area of the restaurant, barely visible in weak moonlight through the windows. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the nearly unlit restaurant hall, closed bedroom door, window moonlight, and quiet nighttime state. Exclude the earlier rear-facing door inspection and show the woman alone in the hall.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 6,
   "C": 4,
   "A": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "전신 샷 지시를 어기고 프레임이 잘렸으나, 배경 레퍼런스를 완벽히 일치시키고 어깨를 치켜올린 불안한 자세를 가장 잘 구현했습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "지정된 배경과 소품은 유지했지만 전신 프레이밍을 위반했으며, 텍스트에 명시된 어깨를 치켜올리는 동작 연기가 누락되었습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "전신 샷은 충족했으나 레퍼런스에 없는 다수의 창문과 테이블을 임의로 생성하여 공간 설정(Location lock)을 크게 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S24sh4_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트에서 '전신' 샷을 요구했으나, 캐릭터의 무릎 아래가 프레임에서 잘린 채 렌더링되었습니다.",
     "fix_en": "Zoom out the camera framing to include the character's feet and lower legs, making it a full-body shot.",
     "needs_regeneration": false,
     "unfixable": true
    }
   ]
  },
  "fix_skipped": true,
  "fix_skip_reason": "all_issues_unfixable",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S24sh4"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S25sh1::bgfirst_bg": {
  "input_fingerprint": "7737cf5408c56919",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 고산들이 어둠 속 기괴하게 나무 기둥이 덧대어진 거대한 자연동굴 입구 앞에 멈춰 선 뒷모습.\n\nLOCATION (lock): Before the timber-braced mouth of a large natural cave deep in the nighttime forest.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Begin static at eye level behind 고산들, holding him at medium distance with the large cave entrance centered directly ahead on the same axis. His halted back occupies the lower middle of the frame, while the reinforced opening and the unreadable darkness beyond retain greater visual weight above him.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 고산들 in the lower-center of the frame, midground, looks toward large cave entrance; large cave entrance in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 거대한 자연동굴 입구 (입구가 나무 기둥으로 덧대어져 있음) — The opening faces the camera and reveals only the dark interior beyond; used as Primary background anchor directly beyond 고산들, presenting the destination he cannot yet enter; 입구에 덧댄 나무 기둥 (동굴 입구를 보강하고 있음) — Their forward and side faces form a constructed border around the natural opening; used as Creates irregular vertical framing around the cave opening; 동굴 내부 (끝을 알 수 없이 깜깜함); used as Functions as an unreadable void past the entrance.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 고산들이 어둠 속 기괴하게 나무 기둥이 덧대어진 거대한 자연동굴 입구 앞에 멈춰 선 뒷모습.\n\nLOCATION (lock): Before the timber-braced mouth of a large natural cave deep in the nighttime forest.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Begin static at eye level behind 고산들, holding him at medium distance with the large cave entrance centered directly ahead on the same axis. His halted back occupies the lower middle of the frame, while the reinforced opening and the unreadable darkness beyond retain greater visual weight above him.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 고산들 in the lower-center of the frame, midground, looks toward large cave entrance; large cave entrance in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 거대한 자연동굴 입구 (입구가 나무 기둥으로 덧대어져 있음) — The opening faces the camera and reveals only the dark interior beyond; used as Primary background anchor directly beyond 고산들, presenting the destination he cannot yet enter; 입구에 덧댄 나무 기둥 (동굴 입구를 보강하고 있음) — Their forward and side faces form a constructed border around the natural opening; used as Creates irregular vertical framing around the cave opening; 동굴 내부 (끝을 알 수 없이 깜깜함); used as Functions as an unreadable void past the entrance.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S25sh1__bgfirst_bg.png",
  "asset_id": "b458da58-bf95-4462-8435-682a3a9aa219",
  "input_asset_ids": [
   "f4f11369-ef12-455d-887f-536cb3b30621",
   "867587f1-f05d-4f79-8306-85b04124926b"
  ]
 },
 "S25sh1": {
  "input_fingerprint": "05d85467fda51845",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 고산들이 어둠 속 기괴하게 나무 기둥이 덧대어진 거대한 자연동굴 입구 앞에 멈춰 선 뒷모습.\n\nLOCATION (lock): Before the timber-braced mouth of a large natural cave deep in the nighttime forest. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 고산들이 어둠 속 기괴하게 나무 기둥이 덧대어진 거대한 자연동굴 입구 앞에 멈춰 선 뒷모습.\n\nLOCATION (lock): Before the timber-braced mouth of a large natural cave deep in the nighttime forest. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 고산들이 어둠 속 기괴하게 나무 기둥이 덧대어진 거대한 자연동굴 입구 앞에 멈춰 선 뒷모습.\n\nLOCATION (lock): Before the timber-braced mouth of a large natural cave deep in the nighttime forest. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S25sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S25sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L18B06.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "프롬프트에 명시된 '고산들(복수)'이 동굴 입구 앞에 서 있는 뒷모습을 정확하게 연출했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "지정된 장소와 뒷모습은 구현했으나, '고산들'이라는 복수 인물 지시와 달리 단 한 명만 등장해 감점되었습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "프롬프트에 명시된 '고산들(복수)'이 동굴 입구 앞에 서 있는 뒷모습을 정확하게 연출했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "지정된 장소와 뒷모습은 구현했으나, '고산들'이라는 복수 인물 지시와 달리 단 한 명만 등장해 감점되었습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시문에 명시된 복수형('고산들')에 맞게 여러 인물의 뒷모습을 배치하였고, 레퍼런스의 장소와 야간 분위기를 충실히 재현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "지정된 장소와 야간 시간대, 뒷모습 구도는 잘 표현했으나, 복수형 지시어('고산들')와 달리 단 한 명만 렌더링하여 우선순위에서 밀립니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시문에 명시된 복수형('고산들')에 맞게 여러 인물의 뒷모습을 배치하였고, 레퍼런스의 장소와 야간 분위기를 충실히 재현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "지정된 장소와 야간 시간대, 뒷모습 구도는 잘 표현했으나, 복수형 지시어('고산들')와 달리 단 한 명만 렌더링하여 우선순위에서 밀립니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 8,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 8,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "프롬프트에 명시된 '고산들(복수)'이 동굴 입구 앞에 서 있는 뒷모습을 정확하게 연출했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "지정된 장소와 뒷모습은 구현했으나, '고산들'이라는 복수 인물 지시와 달리 단 한 명만 등장해 감점되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L18B06.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "중앙 인물의 재킷 오른쪽 어깨 뒷부분에 영문 텍스트 로고가 있습니다.",
     "fix_en": "Remove the text logo from the back of the jacket, leaving only plain fabric.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the text logo from the back of the jacket, leaving only plain fabric.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S25sh1__bgfirst_bg.png",
   "bg_asset_id": "b458da58-bf95-4462-8435-682a3a9aa219",
   "bg_record_key": "S25sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S26sh3::bgfirst_bg": {
  "input_fingerprint": "05c8e6d04331ba80",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 김지후(강수리영)가 허름한 창고 문에 귀를 바짝 대고 숨을 죽인 측면.\n\nLOCATION (lock): In the restaurant's rear yard, directly outside a shabby outbuilding with light leaking through its door.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at head height in a close side profile parallel to the warehouse door, with 김지후(강수리영)'s ear pressed to its plane and her face placed slightly off-center. Her neck cranes forward and one hand braces lightly near the door while the thin line of escaping light remains visible beside her profile.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 허름한 창고 문 (문 밖으로 불빛이 새어나오고 있음) — Its outer face is seen nearly edge-on beside 김지후(강수리영)'s profile; used as The camera runs parallel to its plane, making the contact between ear and door immediately readable; 반월 식당 뒤뜰 (밤); used as Appears only as limited context beyond the tight listening composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 김지후(강수리영)가 허름한 창고 문에 귀를 바짝 대고 숨을 죽인 측면.\n\nLOCATION (lock): In the restaurant's rear yard, directly outside a shabby outbuilding with light leaking through its door.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at head height in a close side profile parallel to the warehouse door, with 김지후(강수리영)'s ear pressed to its plane and her face placed slightly off-center. Her neck cranes forward and one hand braces lightly near the door while the thin line of escaping light remains visible beside her profile.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 허름한 창고 문 (문 밖으로 불빛이 새어나오고 있음) — Its outer face is seen nearly edge-on beside 김지후(강수리영)'s profile; used as The camera runs parallel to its plane, making the contact between ear and door immediately readable; 반월 식당 뒤뜰 (밤); used as Appears only as limited context beyond the tight listening composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S26sh3__bgfirst_bg.png",
  "asset_id": "e998067c-cbe3-4112-9976-1af2c16ff005",
  "input_asset_ids": [
   "1233061f-8753-4056-ba6e-9812cdc53617",
   "5b6b1317-e8ac-4ab9-9e8e-7b89346574a4",
   "c55a139c-93b6-44f8-89bb-912f4fb6fc2c"
  ]
 },
 "S26sh3": {
  "input_fingerprint": "ec4623db50091004",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 허름한 창고 문에 귀를 바짝 대고 숨을 죽인 측면.\n\nLOCATION (lock): In the restaurant's rear yard, directly outside a shabby outbuilding with light leaking through its door. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 허름한 창고 문에 귀를 바짝 대고 숨을 죽인 측면.\n\nLOCATION (lock): In the restaurant's rear yard, directly outside a shabby outbuilding with light leaking through its door. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 허름한 창고 문에 귀를 바짝 대고 숨을 죽인 측면.\n\nLOCATION (lock): In the restaurant's rear yard, directly outside a shabby outbuilding with light leaking through its door. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S26sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S26sh3.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L22B02.png"
    },
    {
     "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_cave_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "측면 구도와 지정된 펜던트 소품을 정확히 구현했으나, 귀를 문에 완전히 밀착한 행동과 우선순위가 높은 동굴 구조물 레퍼런스는 반영하지 못했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "필수 소품인 펜던트가 누락되었고, 원본 위치 사진과 달리 외부 전구가 켜져 조명 무드를 해쳤으며 동굴 구조물 역시 누락되었습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "측면 구도와 지정된 펜던트 소품을 정확히 구현했으나, 귀를 문에 완전히 밀착한 행동과 우선순위가 높은 동굴 구조물 레퍼런스는 반영하지 못했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "필수 소품인 펜던트가 누락되었고, 원본 위치 사진과 달리 외부 전구가 켜져 조명 무드를 해쳤으며 동굴 구조물 역시 누락되었습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 구조물(동굴) 디자인을 누락한 아쉬움이 있으나, 측면 앵글, 문에 다가간 동작, 그리고 뿔 펜던트 묘사에서 지시문을 훨씬 충실히 따랐습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "측면 클로즈업 대신 전신 풀샷을 렌더링했으며, 귀를 문에 대는 핵심 동작과 뿔 펜던트가 모두 누락되었습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 구조물(동굴) 디자인을 누락한 아쉬움이 있으나, 측면 앵글, 문에 다가간 동작, 그리고 뿔 펜던트 묘사에서 지시문을 훨씬 충실히 따랐습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "측면 클로즈업 대신 전신 풀샷을 렌더링했으며, 귀를 문에 대는 핵심 동작과 뿔 펜던트가 모두 누락되었습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 13,
     "B": 8
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 13,
   "B": 8
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "측면 구도와 지정된 펜던트 소품을 정확히 구현했으나, 귀를 문에 완전히 밀착한 행동과 우선순위가 높은 동굴 구조물 레퍼런스는 반영하지 못했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "필수 소품인 펜던트가 누락되었고, 원본 위치 사진과 달리 외부 전구가 켜져 조명 무드를 해쳤으며 동굴 구조물 역시 누락되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L22B02.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_cave_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지문은 '문에 귀를 바짝 대고' 듣는 동작을 명시했으나, 귀를 대지 않고 문틈을 눈으로 들여다보고 있음.",
     "fix_en": "Turn the character's head to press her right ear flat against the door.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "'숨을 죽인' 상태라는 묘사와 달리 캐릭터의 입이 열려 있음.",
     "fix_en": "Close the character's mouth tightly.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Turn the character's head to press her right ear flat against the door.\n- Close the character's mouth tightly.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S26sh3__bgfirst_bg.png",
   "bg_asset_id": "e998067c-cbe3-4112-9976-1af2c16ff005",
   "bg_record_key": "S26sh3::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate",
   "seed_attached": true
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S26sh4": {
  "input_fingerprint": "dd5b3de6fac6f346",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 살짝 열린 창고 문틈 사이로 한쪽 눈을 바짝 댄 채 안을 응시하는 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): At the closed outbuilding door in the restaurant's rear yard, peering through its narrow gap. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the rough shed door, narrow glowing gap, dark backyard, and candlelight leaking from within. Exclude the ear-to-door pose and move to the required eye-at-gap close framing.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 3,
   "C": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "얼굴 클로즈업과 문틈에 눈을 바짝 댄 샷의 구도를 가장 완벽하게 구현했으며, 레퍼런스의 조명과 디테일도 훌륭하게 유지했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "프레이밍이 요구된 클로즈업보다 넓고, 문틈 사이에 양초가 물리적으로 불가능하게 배치되어 심각한 연출 오류를 범했습니다."
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "요구된 얼굴 클로즈업과 행동을 잘 표현했으나, A 후보에 비해 문틈에서 새어 나오는 빛의 질감과 조명이 다소 인위적입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S26sh3_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트는 한쪽 눈을 문틈에 바짝 댄 상태를 요구하나, 사진 속 인물은 얼굴을 뒤로 물린 채 비스듬히 문틈을 바라보고 있습니다.",
     "fix_en": "Turn the character's head and reposition it to press one eye directly against the glowing gap.",
     "needs_regeneration": false,
     "unfixable": true
    }
   ]
  },
  "fix_skipped": true,
  "fix_skip_reason": "all_issues_unfixable",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S26sh3"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S27sh2": {
  "input_fingerprint": "216a93805bef8d6d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 일렁이는 촛불들 사이로, 피가 가득 찬 사발과 기괴한 검은 염소 머리가 놓인 제단의 전경.\n\nLOCATION (lock): Inside the dark rear-yard outbuilding, where multiple candles illuminate a ritual altar. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The altar remains set with the severed black goat head, a bowl full of dark-red blood, burning candles, incense, and the long antler-tipped spear. Jihu still wears her matching antler-bone pendant.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 6,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지시된 제단의 전경이 잘 구현되었으며, 특히 뿔 달린 창이 제단 위에 정확히 배치되어 프롬프트 요구사항을 가장 잘 충족함."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "제단의 주요 요소들은 잘 나타났으나, 뿔 달린 창이 제단이 아닌 우측 벽에 배치되어 위치 정확도가 약간 떨어짐."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "전반적인 분위기와 제단의 형태는 우수하나, 프롬프트에서 요구한 '뿔 달린 긴 창'이 완전히 누락됨."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L23B03.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_restaurant_property_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "제단에 있는 창의 끝부분이 사슴 뿔이 아닌 금속 재질의 창촉으로 그려져 있습니다.",
     "fix_en": "Replace the metal blade at the top of the vertical spear with a pointed antler tip.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the metal blade at the top of the vertical spear with a pointed antler tip.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트+seed만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_bypass:bg_only"
 },
 "S27sh5": {
  "input_fingerprint": "4e81791763adb268",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 넋을 잃고 제단을 바라보는 등 뒤로, 어두운 천장 구석에 매달린 기괴한 인간형 실루엣이 슬며시 드리워진 찰나.\n\nLOCATION (lock): Within the candlelit outbuilding, beneath a dark upper corner where the wall meets the ceiling. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the candlelit shed interior, blood bowl, goat head, altar, smoke, and eerie decorations. Exclude any earlier empty altar-only composition and add the woman with the ceiling silhouette behind her.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The goat head, blood-filled bowl, candles, incense, and long antler-tipped spear remain arranged on the altar. Jihu still wears her matching antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "'실루엣'을 어둠 속에 매달린 실제 인물로 연출하여, 인물을 단순한 그림자가 아닌 온전한 인간으로 묘사하라는 실사화 지침을 훌륭히 준수했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "캐릭터를 벽에 비친 2D 그림자로 묘사하여, 어떤 경우에도 인물을 단순한 실루엣으로 축소하지 말라는 핵심 지침을 위반했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "천장에 매달린 인물이 중력을 거슬러 수평으로 돌출된 형태로 연출되어, 물리적으로 불가능한 구도(Hard Violation)에 해당합니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S27sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "샷 텍스트에 따르면 기괴한 형체가 제단을 바라보는 지후의 '등 뒤로' 매달려 있어야 하지만, 현재 지후 앞쪽인 배경의 제단 위쪽 구석에 매달려 있습니다.",
     "fix_en": "Remove the hanging body from the background corner above the altar and insert a hanging humanoid figure in the upper foreground so it is positioned behind the woman's back.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "참고 이미지의 제단 위 염소 머리는 뿔이 2개이지만, 생성된 이미지에서는 양옆으로 가로지르는 2개의 뿔이 잘못 추가되어 총 4개의 뿔이 있습니다.",
     "fix_en": "Remove the two extra horizontal horns protruding from the sides of the goat's head, leaving only the two main upward-curving horns.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the two extra horizontal horns protruding from the sides of the goat's head, leaving only the two main upward-curving horns.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S27sh2"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S28sh2::bgfirst_bg": {
  "input_fingerprint": "df14422d36ab12a2",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 가로등 하나 없이 칠흑같이 어두운 숲길 입구를 바라보며 두려움에 굳은 채 마른침을 삼키며 목울대가 솟아오른 순간의 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): In the restaurant's rear yard, looking toward the unlit village path leading into the forest.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Derived from the end of the shallow arc, the camera settles at face height just off Kim Ji-hoo's frontal axis and completes a restrained dolly-in to a close framing. Her rigid face occupies the upper frame while her throat remains visible in the lower third, catching the exact rise of the swallow without losing her fearful stare toward the forest entrance off-screen.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 숲길 입구 (가로등이 없고 매우 어두움) — The entrance recedes away from the camera behind the side of her face; used as The unseen destination of her fixed gaze, with only its near edge retained behind her as sparse spatial context.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 가로등 하나 없이 칠흑같이 어두운 숲길 입구를 바라보며 두려움에 굳은 채 마른침을 삼키며 목울대가 솟아오른 순간의 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): In the restaurant's rear yard, looking toward the unlit village path leading into the forest.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Derived from the end of the shallow arc, the camera settles at face height just off Kim Ji-hoo's frontal axis and completes a restrained dolly-in to a close framing. Her rigid face occupies the upper frame while her throat remains visible in the lower third, catching the exact rise of the swallow without losing her fearful stare toward the forest entrance off-screen.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 숲길 입구 (가로등이 없고 매우 어두움) — The entrance recedes away from the camera behind the side of her face; used as The unseen destination of her fixed gaze, with only its near edge retained behind her as sparse spatial context.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S28sh2__bgfirst_bg.png",
  "asset_id": "290a16f3-7f7c-46d1-8d3a-34a1aa119707",
  "input_asset_ids": [
   "f47dd3af-a525-4020-9937-f09054f464f0",
   "eb92a04d-eb6b-4427-99de-f49275f41de0"
  ]
 },
 "S28sh2": {
  "input_fingerprint": "6852cb7ba68eb546",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 가로등 하나 없이 칠흑같이 어두운 숲길 입구를 바라보며 두려움에 굳은 채 마른침을 삼키며 목울대가 솟아오른 순간의 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): In the restaurant's rear yard, looking toward the unlit village path leading into the forest. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 가로등 하나 없이 칠흑같이 어두운 숲길 입구를 바라보며 두려움에 굳은 채 마른침을 삼키며 목울대가 솟아오른 순간의 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): In the restaurant's rear yard, looking toward the unlit village path leading into the forest. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 가로등 하나 없이 칠흑같이 어두운 숲길 입구를 바라보며 두려움에 굳은 채 마른침을 삼키며 목울대가 솟아오른 순간의 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): In the restaurant's rear yard, looking toward the unlit village path leading into the forest. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S28sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S28sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L24B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "요구된 클로즈업 프레이밍에 가깝고, 두려움에 굳어 마른침을 삼키며 목울대가 도드라지는 연기와 표정을 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "배경과 인물은 일치하나 샷이 다소 넓고, 침을 삼키는 목의 움직임과 두려운 감정 표현이 누락된 채 정면만 응시하고 있습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "요구된 클로즈업 프레이밍에 가깝고, 두려움에 굳어 마른침을 삼키며 목울대가 도드라지는 연기와 표정을 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "배경과 인물은 일치하나 샷이 다소 넓고, 침을 삼키는 목의 움직임과 두려운 감정 표현이 누락된 채 정면만 응시하고 있습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지시된 얼굴 클로즈업과 두려움에 마른침을 삼키며 목울대가 솟아오른 디테일을 매우 충실하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "클로즈업 지시를 어기고 프레임을 넓혀 펜던트를 보여주었으며, 요구된 목울대와 표정 디테일이 부족합니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지시된 얼굴 클로즈업과 두려움에 마른침을 삼키며 목울대가 솟아오른 디테일을 매우 충실하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "클로즈업 지시를 어기고 프레임을 넓혀 펜던트를 보여주었으며, 요구된 목울대와 표정 디테일이 부족합니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 9,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 9,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "요구된 클로즈업 프레이밍에 가깝고, 두려움에 굳어 마른침을 삼키며 목울대가 도드라지는 연기와 표정을 훌륭하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "배경과 인물은 일치하나 샷이 다소 넓고, 침을 삼키는 목의 움직임과 두려운 감정 표현이 누락된 채 정면만 응시하고 있습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L24B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "인물이 화면 앞쪽(왼쪽 밖)을 바라보고 있어, 그녀의 뒤쪽 배경에 위치한 숲길 입구를 응시해야 한다는 지문의 상황과 공간적으로 모순됩니다.",
     "fix_en": "Turn the character's head and eyes to look back over her right shoulder directly toward the dark path in the background.",
     "unfixable": true
    },
    {
     "issue_ko": "침을 삼키는 묘사가 과장되어 목 부분에 깊은 세로 주름과 지나치게 굵은 근육이 생성되어 젊은 여성의 자연스러운 해부학적 구조와 어긋납니다.",
     "fix_en": "Soften and smooth the neck anatomy to look natural for a young woman, reducing the exaggerated vertical muscles and deep wrinkles while keeping a subtle hint of swallowing.",
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Soften and smooth the neck anatomy to look natural for a young woman, reducing the exaggerated vertical muscles and deep wrinkles while keeping a subtle hint of swallowing.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S28sh2__bgfirst_bg.png",
   "bg_asset_id": "290a16f3-7f7c-46d1-8d3a-34a1aa119707",
   "bg_record_key": "S28sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S29sh1::bgfirst_bg": {
  "input_fingerprint": "a087a5b20c105593",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 해송 숲길에서 김지후(강수리영)가 켜진 핸드폰 플래시 불빛을 앞으로 뻗어 비춘 채 긴장한 표정으로 서 있는 전신.\n\nLOCATION (lock): On a narrow, unlit path inside the nighttime pine forest, illuminated only by a phone flashlight.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level on the path centerline, the static camera holds a wide full-body view of Kim Ji-hoo facing into the route. She is centered with her lit phone extended into the upper-middle foreground, while the narrow path and pine forest remain readable around her before any retreat toward cover begins.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 좁은 해송 숲길 (밤의 어두운 숲길) — The route extends from the lower foreground behind Kim Ji-hoo into the distant center; used as Carries the phone beam forward and establishes the confined route around the full figure; 핸드폰 (플래시가 켜짐) — Its flash side points away from Kim Ji-hoo and along the path, with the body seen obliquely by camera; used as Held forward as the principal visible illumination and directional cue.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 해송 숲길에서 김지후(강수리영)가 켜진 핸드폰 플래시 불빛을 앞으로 뻗어 비춘 채 긴장한 표정으로 서 있는 전신.\n\nLOCATION (lock): On a narrow, unlit path inside the nighttime pine forest, illuminated only by a phone flashlight.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level on the path centerline, the static camera holds a wide full-body view of Kim Ji-hoo facing into the route. She is centered with her lit phone extended into the upper-middle foreground, while the narrow path and pine forest remain readable around her before any retreat toward cover begins.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 좁은 해송 숲길 (밤의 어두운 숲길) — The route extends from the lower foreground behind Kim Ji-hoo into the distant center; used as Carries the phone beam forward and establishes the confined route around the full figure; 핸드폰 (플래시가 켜짐) — Its flash side points away from Kim Ji-hoo and along the path, with the body seen obliquely by camera; used as Held forward as the principal visible illumination and directional cue.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S29sh1__bgfirst_bg.png",
  "asset_id": "cb5b275f-54ee-4886-aa9c-61f94662990d",
  "input_asset_ids": [
   "55d7a2f5-3301-443d-a7a1-6034b0ea627e",
   "947f1562-e8ef-43fc-ad1a-3ca896912adb"
  ]
 },
 "S29sh1": {
  "input_fingerprint": "adb334071ea71f41",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 해송 숲길에서 김지후(강수리영)가 켜진 핸드폰 플래시 불빛을 앞으로 뻗어 비춘 채 긴장한 표정으로 서 있는 전신.\n\nLOCATION (lock): On a narrow, unlit path inside the nighttime pine forest, illuminated only by a phone flashlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and carries her phone, using its flashlight.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 해송 숲길에서 김지후(강수리영)가 켜진 핸드폰 플래시 불빛을 앞으로 뻗어 비춘 채 긴장한 표정으로 서 있는 전신.\n\nLOCATION (lock): On a narrow, unlit path inside the nighttime pine forest, illuminated only by a phone flashlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and carries her phone, using its flashlight.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 해송 숲길에서 김지후(강수리영)가 켜진 핸드폰 플래시 불빛을 앞으로 뻗어 비춘 채 긴장한 표정으로 서 있는 전신.\n\nLOCATION (lock): On a narrow, unlit path inside the nighttime pine forest, illuminated only by a phone flashlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and carries her phone, using its flashlight.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S29sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S29sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L23B06.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "핸드폰 플래시를 '앞으로 뻗어 비춘 채' 서 있는 전신 묘사와 인상착의가 지시문에 정확히 부합합니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "전신 샷과 배경은 훌륭하나, 팔을 앞으로 뻗은 자세가 다소 소극적으로 표현되었습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "핸드폰 플래시를 '앞으로 뻗어 비춘 채' 서 있는 전신 묘사와 인상착의가 지시문에 정확히 부합합니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "전신 샷과 배경은 훌륭하나, 팔을 앞으로 뻗은 자세가 다소 소극적으로 표현되었습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지시문의 '앞으로 뻗어 비춘 채'를 팔을 뻗은 포즈와 조명 방향으로 정확히 구현했으며 전신 샷의 요구사항도 잘 충족함."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "전신 샷과 캐릭터 설정은 일치하나, 팔을 뻗어 비추는 동작('앞으로 뻗어')이 다소 소극적으로 표현되어 아쉬움."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지시문의 '앞으로 뻗어 비춘 채'를 팔을 뻗은 포즈와 조명 방향으로 정확히 구현했으며 전신 샷의 요구사항도 잘 충족함."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "전신 샷과 캐릭터 설정은 일치하나, 팔을 뻗어 비추는 동작('앞으로 뻗어')이 다소 소극적으로 표현되어 아쉬움."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 11
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 15,
   "B": 11
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "핸드폰 플래시를 '앞으로 뻗어 비춘 채' 서 있는 전신 묘사와 인상착의가 지시문에 정확히 부합합니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "전신 샷과 배경은 훌륭하나, 팔을 앞으로 뻗은 자세가 다소 소극적으로 표현되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L23B06.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "핸드폰을 쥔 왼손이 해부학적으로 잘못되었습니다(손등이 보일 때 엄지손가락이 왼쪽에 위치함).",
     "fix_en": "Redraw the left hand holding the phone as an anatomically correct left hand, placing the thumb on the right side.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "목걸이의 가죽 줄이 누락되었으며, 펜던트가 뼈 재질이 아닌 금빛으로 잘못 표현되었습니다.",
     "fix_en": "Add a dark leather cord around her neck and change the pendant to a matte bone color to match the reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the left hand holding the phone as an anatomically correct left hand, placing the thumb on the right side.\n- Add a dark leather cord around her neck and change the pendant to a matte bone color to match the reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S29sh1__bgfirst_bg.png",
   "bg_asset_id": "cb5b275f-54ee-4886-aa9c-61f94662990d",
   "bg_record_key": "S29sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S29sh3": {
  "input_fingerprint": "f506d71b1dc5d02a",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 수풀에 몸을 낮춘 김지후(강수리영)의 입을 뒤에서 나타난 두툼한 손이 강하게 꽉 틀어막은 순간.\n\nLOCATION (lock): Behind low vegetation beside the dark pine-forest path. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the narrow pine path, dense undergrowth, darkness, and phone-flash lighting. Exclude the earlier upright walking pose and show the woman crouched with her mouth covered.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and retains her phone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 4,
   "A": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지시된 동작(뒤에서 입을 틀어막음)과 인물, 소품의 디테일을 해부학적 오류 없이 가장 정확하게 구현했습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "뒤에서 나타난 손이 입을 강하게 틀어막는 핵심 동작을 연출하지 않고 어깨에만 손을 올려 샷의 주요 상황을 놓쳤습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "동작은 연출했으나, 습격자의 팔 방향과 손(왼팔 위치에서 뻗어나온 오른손)의 해부학적 구조가 물리적으로 불가능한 치명적 오류가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S29sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "여자의 입을 막고 있는 손의 손가락 개수가 너무 많습니다 (얼굴을 덮고 있는 손가락만 5개임).",
     "fix_en": "Remove the extra bottom finger from the hand covering the woman's mouth so that only four fingers rest across her face.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the extra bottom finger from the hand covering the woman's mouth so that only four fingers rest across her face.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S29sh1"
  }
 },
 "S29sh5": {
  "input_fingerprint": "c7e5e1b2bff19d30",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김여상이 김지후(강수리영)의 뒤에 밀착한 채 입술 위로 손가락을 올려 '쉿' 동작을 취하는 찰나.\n\nLOCATION (lock): In cover among the undergrowth at the edge of the nighttime forest path. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the dark pine forest, low brush, and concealed roadside position. Exclude the unseen hand-only attack moment and show the man behind her making the hush gesture.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and retains her phone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 3,
   "A": 2
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "해부학적 오류 없이 남자가 여자의 뒤에 밀착해 '쉿' 하는 동작과 모든 프롬프트 설정을 정확히 구현했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "입술에 닿은 손의 소매가 여자의 스웨터와 동일해 세 번째 팔이 생긴 듯한 신체 구조적 하드 위반이 있습니다."
   },
   {
    "label": "A",
    "score": 2,
    "verdict_ko": "가슴과 입술 주변에 출처가 불분명한 손들이 얽혀 있어 심각한 해부학적 하드 위반이 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S29sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지후가 스마트폰을 쥐고 있는 오른손에 손가락이 너무 많습니다. 폰 가장자리에 뻗은 검지 외에도 폰 아래를 감싸는 손가락이 4개나 있어 총 6개의 손가락이 보입니다.",
     "fix_en": "Remove one of the extra fingers wrapping underneath the phone in Jihu's right hand so that she has a normal hand with exactly five digits.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove one of the extra fingers wrapping underneath the phone in Jihu's right hand so that she has a normal hand with exactly five digits.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S29sh3"
  }
 },
 "S30sh2": {
  "input_fingerprint": "f7aeecb2b20e5ef8",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 포댓자루 겉면에 검붉은 액체가 묻어 있고, 안쪽에서 가해진 힘으로 천이 바깥으로 팽팽하게 불룩 튀어나온 상태의 클로즈업.\n\nLOCATION (lock): Deep in the pathless nighttime forest, on the route followed by the youths carrying sacks. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The two sacks remain occupied by struggling captives and continue leaking dark-red liquid.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 5,
   "A": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "안쪽에서 가해진 힘으로 팽팽하게 불룩해진 포댓자루의 상태를 가장 잘 구현했으나, 지시와 달리 인물의 일부가 프레임에 노출되었습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "인물을 배제하라는 지시는 지켰으나, 내부의 힘으로 천이 팽팽하게 튀어나온 핵심적인 묘사가 부족합니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "불룩 튀어나온 형태는 잘 살렸으나, 포댓자루 중앙에 텍스트 형태의 픽셀 아티팩트가 누출되어 중대한 규정 위반입니다."
   }
  ],
  "refs": [],
  "critique": {
   "issues": [
    {
     "issue_ko": "화면 왼쪽 가장자리와 포댓자루 뒤편에 사람의 신체(팔, 다리, 옷)가 보입니다. 지문에 사람이 등장하지 않아야 한다고 명시되어 있습니다.",
     "fix_en": "Remove the person's arm, torso, and leg visible on the left side and behind the sack, replacing them completely with the dark forest background."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the person's arm, torso, and leg visible on the left side and behind the sack, replacing them completely with the dark forest background.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S30sh4": {
  "input_fingerprint": "2c10a34b30a21cd0",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 발각될 위기에 처한 김지후(강수리영)와 김여상이 수풀 뒤 바닥에 납작 엎드려 숨을 죽인 상태.\n\nLOCATION (lock): Flat behind low brush in the deep forest, just off the youths' route. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and carries her phone while she and Yeo-sang hide prone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상); 김여상 (한국인 40대 중반 남성, 평범한 중년 얼굴, 짧은 검은 머리, 소박한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 5,
   "B": 7,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "엎드린 자세와 주변 환경은 잘 표현되었으나, 김여상의 의상이 레퍼런스와 전혀 다르게 묘사되어 감점됨."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지정된 자세, 배경, 자루의 형태 및 주요 의상을 가장 잘 재현했으나 김여상의 모자가 누락됨."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "매달린 자루가 가로로 왜곡되어 있고 스마트폰이 엉덩이에 부자연스럽게 붙어있는 물리적 오류가 있음."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S30sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   },
   {
    "label": "CHARACTER REFERENCE — 김여상: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:888076>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지후의 뿔 펜던트가 목에 연결된 줄 없이 바닥에 떨어져 있습니다.",
     "fix_en": "Add a dark necklace cord connecting the antler-bone pendant on the ground to Jihu's neck so it is clear she is wearing it.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "화면 오른쪽 수풀에 배경 레퍼런스에는 없는 주황색 불빛들이 떠 있습니다.",
     "fix_en": "Remove the floating glowing orange dots from the bushes on the right to match the dark lighting mood of the location reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add a dark necklace cord connecting the antler-bone pendant on the ground to Jihu's neck so it is clear she is wearing it.\n- Remove the floating glowing orange dots from the bushes on the right to match the dark lighting mood of the location reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S30sh2"
  }
 },
 "S31sh2::bgfirst_bg": {
  "input_fingerprint": "56048c1ef0afa94b",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 꿈틀거리는 포댓자루를 메고 한 발은 딛고 다른 발은 허공에 든 채 검은 염소 떼 사이를 가로지르는 금월도 마을 청년들의 무심한 표정.\n\nLOCATION (lock): Inside the crowded pen of a forest goat shed, along the narrow passage through the animals.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Tracking at chest height alongside the route, the camera holds a wide lateral view of the village youths threading through the herd. Their bodies cross the middle band with alternating lifted feet, while the struggling sack remains readable over a carrier's shoulder and black goats occupy the foreground and background without blocking the procession.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 검은 염소 떼 (수십 마리가 축사 안에 몰려있음) — Individual goats face varied directions on both sides of the youths' lateral route; used as Surrounds and intermittently layers across the procession, establishing the youths' route through the enclosure; 축사 내부 통로 (마을 청년들이 통과 중) — The route runs laterally across the frame between the gathered goats; used as Provides the constrained passage followed by the lateral tracking camera; 포댓자루 (안의 무언가가 사력을 다해 꿈틀거림) — Its long side is visible over a youth's shoulder and deforms outward as the contents struggle; used as Registers the violent internal movement ignored by the carriers.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 꿈틀거리는 포댓자루를 메고 한 발은 딛고 다른 발은 허공에 든 채 검은 염소 떼 사이를 가로지르는 금월도 마을 청년들의 무심한 표정.\n\nLOCATION (lock): Inside the crowded pen of a forest goat shed, along the narrow passage through the animals.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Tracking at chest height alongside the route, the camera holds a wide lateral view of the village youths threading through the herd. Their bodies cross the middle band with alternating lifted feet, while the struggling sack remains readable over a carrier's shoulder and black goats occupy the foreground and background without blocking the procession.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 검은 염소 떼 (수십 마리가 축사 안에 몰려있음) — Individual goats face varied directions on both sides of the youths' lateral route; used as Surrounds and intermittently layers across the procession, establishing the youths' route through the enclosure; 축사 내부 통로 (마을 청년들이 통과 중) — The route runs laterally across the frame between the gathered goats; used as Provides the constrained passage followed by the lateral tracking camera; 포댓자루 (안의 무언가가 사력을 다해 꿈틀거림) — Its long side is visible over a youth's shoulder and deforms outward as the contents struggle; used as Registers the violent internal movement ignored by the carriers.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S31sh2__bgfirst_bg.png",
  "asset_id": "948ef996-7db9-471f-8213-32945990c3d9",
  "input_asset_ids": [
   "57df3218-af71-4d46-8947-8c0aca1cce73",
   "803cc956-9eff-49c4-b977-3e62ad55fc55"
  ]
 },
 "S31sh2": {
  "input_fingerprint": "fb672bf5b68852b5",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 꿈틀거리는 포댓자루를 메고 한 발은 딛고 다른 발은 허공에 든 채 검은 염소 떼 사이를 가로지르는 금월도 마을 청년들의 무심한 표정.\n\nLOCATION (lock): Inside the crowded pen of a forest goat shed, along the narrow passage through the animals. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The carried sacks still contain struggling captives and are stained with leaking dark-red liquid. Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 청년 (한국인 20~30대 남성, 짧은 검은 머리, 그을린 얼굴, 거친 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 꿈틀거리는 포댓자루를 메고 한 발은 딛고 다른 발은 허공에 든 채 검은 염소 떼 사이를 가로지르는 금월도 마을 청년들의 무심한 표정.\n\nLOCATION (lock): Inside the crowded pen of a forest goat shed, along the narrow passage through the animals. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The carried sacks still contain struggling captives and are stained with leaking dark-red liquid. Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 청년 (한국인 20~30대 남성, 짧은 검은 머리, 그을린 얼굴, 거친 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 꿈틀거리는 포댓자루를 메고 한 발은 딛고 다른 발은 허공에 든 채 검은 염소 떼 사이를 가로지르는 금월도 마을 청년들의 무심한 표정.\n\nLOCATION (lock): Inside the crowded pen of a forest goat shed, along the narrow passage through the animals. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The carried sacks still contain struggling captives and are stained with leaking dark-red liquid. Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 청년 (한국인 20~30대 남성, 짧은 검은 머리, 그을린 얼굴, 거친 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S31sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S31sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 청년: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:756328>"
    },
    {
     "label": "PROP REFERENCE — 피 묻은 대형 포댓자루: the exact object appearing in this shot; match its look, material and wear exactly.",
     "path": "<bytes:1683711>"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 청년: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:756328>"
    },
    {
     "label": "PROP REFERENCE — 피 묻은 대형 포댓자루: the exact object appearing in this shot; match its look, material and wear exactly.",
     "path": "<bytes:1683711>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 인물과 피 묻은 자루 레퍼런스를 정확히 반영했으며, 한 발을 허공에 든 채 걷는 묘사와 무심한 표정을 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "동일한 인물이 화면에 여러 번 복제되어 나타나는 심각한 규정 위반(Hard Violation: duplicated bodies)이 발생했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 인물과 피 묻은 자루 레퍼런스를 정확히 반영했으며, 한 발을 허공에 든 채 걷는 묘사와 무심한 표정을 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "동일한 인물이 화면에 여러 번 복제되어 나타나는 심각한 규정 위반(Hard Violation: duplicated bodies)이 발생했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "지정된 숲속 비석 레퍼런스 배경이 누락되었으나, 인물의 동작, 포댓자루, 염소 떼 등 텍스트의 요구사항을 자연스럽게 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "동일한 인물이 비정상적으로 5명이나 복제되어 일렬로 나타나는 치명적인 다중 신체 오류(Hard Violation)가 발생했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "지정된 숲속 비석 레퍼런스 배경이 누락되었으나, 인물의 동작, 포댓자루, 염소 떼 등 텍스트의 요구사항을 자연스럽게 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "동일한 인물이 비정상적으로 5명이나 복제되어 일렬로 나타나는 치명적인 다중 신체 오류(Hard Violation)가 발생했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 5,
     "B": 13
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 5,
   "B": 13
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지정된 인물과 피 묻은 자루 레퍼런스를 정확히 반영했으며, 한 발을 허공에 든 채 걷는 묘사와 무심한 표정을 훌륭하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 2,
    "verdict_ko": "동일한 인물이 화면에 여러 번 복제되어 나타나는 심각한 규정 위반(Hard Violation: duplicated bodies)이 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 청년: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:756328>"
   },
   {
    "label": "PROP REFERENCE — 피 묻은 대형 포댓자루: the exact object appearing in this shot; match its look, material and wear exactly.",
    "path": "<bytes:1683711>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "배경이 실내 나무 축사로 생성되어, 야외 소나무 숲길과 대형 비석이 있는 지정된 로케이션 레퍼런스 사진의 장소 구조를 완전히 무시했습니다.",
     "fix_en": "Replace the entire indoor wooden barn environment with the outdoor dark pine forest path and stone monument shown in the location reference photograph.",
     "needs_regeneration": true,
     "unfixable": true
    }
   ]
  },
  "fix_skipped": true,
  "fix_skip_reason": "all_issues_unfixable",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S31sh2__bgfirst_bg.png",
   "bg_asset_id": "948ef996-7db9-471f-8213-32945990c3d9",
   "bg_record_key": "S31sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S32sh1::bgfirst_bg": {
  "input_fingerprint": "dc55e3dca69a0146",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 얼기설기 지어진 나무 가건물 밖 어둠 속에서 여상이 벽면에 얼굴을 가까이 대고 안을 들여다보는 고정된 자세.\n\nLOCATION (lock): Outside the rough wooden slaughterhouse hidden in the forest, close against its wall and broken window.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Outside the structure, the static eye-level camera sits close to the wall and looks along it toward Kim Yeo-sang's profile. He occupies one side of the medium frame with his face pressed near the wall, while the receding exterior surface leaves searching space ahead in the direction the later track will follow.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 가건물 외벽 (나무로 얼기설기 지어진 가건물의 외벽) — The near side of the wall runs diagonally away from camera past Kim Yeo-sang; used as Acts as the dominant leading plane, guiding both his search and the next parallel camera move; 가건물 주변 숲속 (밤); used as Holds empty searching space ahead of his profile and establishes the isolated exterior location.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 얼기설기 지어진 나무 가건물 밖 어둠 속에서 여상이 벽면에 얼굴을 가까이 대고 안을 들여다보는 고정된 자세.\n\nLOCATION (lock): Outside the rough wooden slaughterhouse hidden in the forest, close against its wall and broken window.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Outside the structure, the static eye-level camera sits close to the wall and looks along it toward Kim Yeo-sang's profile. He occupies one side of the medium frame with his face pressed near the wall, while the receding exterior surface leaves searching space ahead in the direction the later track will follow.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 가건물 외벽 (나무로 얼기설기 지어진 가건물의 외벽) — The near side of the wall runs diagonally away from camera past Kim Yeo-sang; used as Acts as the dominant leading plane, guiding both his search and the next parallel camera move; 가건물 주변 숲속 (밤); used as Holds empty searching space ahead of his profile and establishes the isolated exterior location.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S32sh1__bgfirst_bg.png",
  "asset_id": "f88b8c04-86e7-4a5c-858f-d37206c23bb7",
  "input_asset_ids": [
   "a83db6b1-23d3-4a61-8277-694e787e4ee5",
   "45402dff-76ad-4136-9658-d7261ccee20f",
   "6434b544-0a16-42ea-a909-412c18868a98"
  ]
 },
 "S32sh1": {
  "input_fingerprint": "2fbeb661e584a959",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 얼기설기 지어진 나무 가건물 밖 어둠 속에서 여상이 벽면에 얼굴을 가까이 대고 안을 들여다보는 고정된 자세.\n\nLOCATION (lock): Outside the rough wooden slaughterhouse hidden in the forest, close against its wall and broken window. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and carries her phone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김여상 (한국인 40대 중반 남성, 평범한 중년 얼굴, 짧은 검은 머리, 소박한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 얼기설기 지어진 나무 가건물 밖 어둠 속에서 여상이 벽면에 얼굴을 가까이 대고 안을 들여다보는 고정된 자세.\n\nLOCATION (lock): Outside the rough wooden slaughterhouse hidden in the forest, close against its wall and broken window. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and carries her phone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김여상 (한국인 40대 중반 남성, 평범한 중년 얼굴, 짧은 검은 머리, 소박한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 얼기설기 지어진 나무 가건물 밖 어둠 속에서 여상이 벽면에 얼굴을 가까이 대고 안을 들여다보는 고정된 자세.\n\nLOCATION (lock): Outside the rough wooden slaughterhouse hidden in the forest, close against its wall and broken window. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her antler-bone pendant and carries her phone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김여상 (한국인 40대 중반 남성, 평범한 중년 얼굴, 짧은 검은 머리, 소박한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S32sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S32sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김여상: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:888076>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L25B03.png"
    },
    {
     "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_slaughter_complex_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김여상: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:888076>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "벽면에 얼굴을 바짝 대고 창문 안을 들여다보는 자세를 완벽히 구현했으며 샷의 분위기와 구조물 묘사도 훌륭합니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "얼굴을 벽에 밀착하지 않아 지시된 자세를 어겼으며, 텍스트에 없는 다른 캐릭터의 소품(펜던트, 폰)을 장착한 오류가 있습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "벽면에 얼굴을 바짝 대고 창문 안을 들여다보는 자세를 완벽히 구현했으며 샷의 분위기와 구조물 묘사도 훌륭합니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "얼굴을 벽에 밀착하지 않아 지시된 자세를 어겼으며, 텍스트에 없는 다른 캐릭터의 소품(펜던트, 폰)을 장착한 오류가 있습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "'벽면에 얼굴을 가까이 대고 안을 들여다보는' 지시문의 구체적인 자세를 정확하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "얼굴을 벽에 밀착하는 자세가 부족하며, 텍스트 상 다른 인물(지후)의 소지품을 잘못 착용했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "'벽면에 얼굴을 가까이 대고 안을 들여다보는' 지시문의 구체적인 자세를 정확하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "얼굴을 벽에 밀착하는 자세가 부족하며, 텍스트 상 다른 인물(지후)의 소지품을 잘못 착용했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 8,
     "B": 15
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 8,
   "B": 15
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "벽면에 얼굴을 바짝 대고 창문 안을 들여다보는 자세를 완벽히 구현했으며 샷의 분위기와 구조물 묘사도 훌륭합니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "얼굴을 벽에 밀착하지 않아 지시된 자세를 어겼으며, 텍스트에 없는 다른 캐릭터의 소품(펜던트, 폰)을 장착한 오류가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L25B03.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_slaughter_complex_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김여상: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:888076>"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S32sh1__bgfirst_bg.png",
   "bg_asset_id": "f88b8c04-86e7-4a5c-858f-d37206c23bb7",
   "bg_record_key": "S32sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate",
   "seed_attached": true
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S32sh10::bgfirst_bg": {
  "input_fingerprint": "0594b0faa220826d",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 한 마을 청년이 허공 높이 무시무시한 도축용 칼을 치켜든 순간.\n\nLOCATION (lock): Inside the dark forest slaughterhouse, among hanging knives, hooks, and blood-catching equipment.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Inside the structure at waist height, the camera completes its dolly-in from a medium distance on an upward angle toward the village youth's raised arm. His braced torso occupies one side of the frame and the slaughter knife reaches the upper opposite quadrant at its highest point, while only the lower edge of the struggling figure remains visible beneath the intended strike.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 도축용 칼 (허공 높이 치켜들림) — The blade is seen obliquely above the youth, with its cutting direction aimed toward the lower frame; used as Creates the upper visual apex and preserves the threat direction from raised blade to the lower-frame victim; 벽에 매달린 다양한 크기의 칼 (벽에 줄줄이 매달림) — Their sides and hanging profiles are visible in a row along the rear wall; used as Forms a ritual-like row behind the attacker without competing with the raised weapon; 걸개 갈고리와 피받이 양동이 (도축 도구로 배치됨) — The hooks hang against the rear area while the buckets remain below them; used as Retains the slaughterhouse function around the attack in the background.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 한 마을 청년이 허공 높이 무시무시한 도축용 칼을 치켜든 순간.\n\nLOCATION (lock): Inside the dark forest slaughterhouse, among hanging knives, hooks, and blood-catching equipment.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Inside the structure at waist height, the camera completes its dolly-in from a medium distance on an upward angle toward the village youth's raised arm. His braced torso occupies one side of the frame and the slaughter knife reaches the upper opposite quadrant at its highest point, while only the lower edge of the struggling figure remains visible beneath the intended strike.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 도축용 칼 (허공 높이 치켜들림) — The blade is seen obliquely above the youth, with its cutting direction aimed toward the lower frame; used as Creates the upper visual apex and preserves the threat direction from raised blade to the lower-frame victim; 벽에 매달린 다양한 크기의 칼 (벽에 줄줄이 매달림) — Their sides and hanging profiles are visible in a row along the rear wall; used as Forms a ritual-like row behind the attacker without competing with the raised weapon; 걸개 갈고리와 피받이 양동이 (도축 도구로 배치됨) — The hooks hang against the rear area while the buckets remain below them; used as Retains the slaughterhouse function around the attack in the background.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S32sh10__bgfirst_bg.png",
  "asset_id": "a6d62c9f-35c2-411f-8de7-68d83d2f8565",
  "input_asset_ids": [
   "8a675677-a545-416e-ab90-20b10e1eb081",
   "45402dff-76ad-4136-9658-d7261ccee20f"
  ]
 },
 "S32sh10": {
  "input_fingerprint": "1f38ff9b38ff98f6",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 한 마을 청년이 허공 높이 무시무시한 도축용 칼을 치켜든 순간.\n\nLOCATION (lock): Inside the dark forest slaughterhouse, among hanging knives, hooks, and blood-catching equipment. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The captive removed from the sack remains on the slaughterhouse floor, struggling for life. Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 청년 (한국인 20~30대 남성, 짧은 검은 머리, 그을린 얼굴, 거친 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 한 마을 청년이 허공 높이 무시무시한 도축용 칼을 치켜든 순간.\n\nLOCATION (lock): Inside the dark forest slaughterhouse, among hanging knives, hooks, and blood-catching equipment. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The captive removed from the sack remains on the slaughterhouse floor, struggling for life. Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 청년 (한국인 20~30대 남성, 짧은 검은 머리, 그을린 얼굴, 거친 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 한 마을 청년이 허공 높이 무시무시한 도축용 칼을 치켜든 순간.\n\nLOCATION (lock): Inside the dark forest slaughterhouse, among hanging knives, hooks, and blood-catching equipment. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The captive removed from the sack remains on the slaughterhouse floor, struggling for life. Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 청년 (한국인 20~30대 남성, 짧은 검은 머리, 그을린 얼굴, 거친 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S32sh10__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S32sh10.png"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 청년: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:756328>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L25B03.png"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 청년: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:756328>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지문에서 가장 핵심적인 '도축용 칼을 치켜든' 행동이 누락된 채 빈 손을 들고 있어 치명적인 오류입니다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "거대한 도축용 칼을 치켜든 행동을 정확히 연출했으며, 뼈 펜던트와 바닥에 쓰러진 인질 등 구체적인 상태 지시까지 충실히 반영했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지문에서 가장 핵심적인 '도축용 칼을 치켜든' 행동이 누락된 채 빈 손을 들고 있어 치명적인 오류입니다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "거대한 도축용 칼을 치켜든 행동을 정확히 연출했으며, 뼈 펜던트와 바닥에 쓰러진 인질 등 구체적인 상태 지시까지 충실히 반영했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "도축용 칼을 치켜든 샷 텍스트의 핵심 행동을 훌륭히 구현했으나, 인물 제한 지침을 어기고 배경 탁자 위에 포로가 등장한 점이 감점 요인입니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "가장 중요한 소품인 도축용 칼이 완전히 누락되어 샷 텍스트가 요구하는 핵심 행동을 전혀 표현하지 못했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "도축용 칼을 치켜든 샷 텍스트의 핵심 행동을 훌륭히 구현했으나, 인물 제한 지침을 어기고 배경 탁자 위에 포로가 등장한 점이 감점 요인입니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "가장 중요한 소품인 도축용 칼이 완전히 누락되어 샷 텍스트가 요구하는 핵심 행동을 전혀 표현하지 못했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "지문에서 가장 핵심적인 '도축용 칼을 치켜든' 행동이 누락된 채 빈 손을 들고 있어 치명적인 오류입니다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "거대한 도축용 칼을 치켜든 행동을 정확히 연출했으며, 뼈 펜던트와 바닥에 쓰러진 인질 등 구체적인 상태 지시까지 충실히 반영했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L25B03.png"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 청년: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:756328>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "청년의 재킷 앞면에 사슴 뿔 펜던트가 달려 있습니다(프롬프트 설정상 지후의 소품임).",
     "fix_en": "Remove the antler-bone pendant from the front of the youth's jacket.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "포로가 도축장 바닥에 있어야 한다는 지시와 달리 배경의 테이블 위에 누워 있습니다.",
     "fix_en": "Move the person in the background from the table down to the floor.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the antler-bone pendant from the front of the youth's jacket.\n- Move the person in the background from the table down to the floor.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S32sh10__bgfirst_bg.png",
   "bg_asset_id": "a6d62c9f-35c2-411f-8de7-68d83d2f8565",
   "bg_record_key": "S32sh10::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S32sh13": {
  "input_fingerprint": "34dbb3ad5add9af7",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 플래시 불빛에 노출된 피범벅 상태의 마을 청년들이 창문 쪽으로 일제히 고개를 돌린 얼굴.\n\nLOCATION (lock): Inside the slaughterhouse facing its broken window, momentarily exposed by a camera flash. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the broken window, rough timber walls, dusty glass, and dark slaughterhouse lighting. Exclude the knife's earlier raised position and show the bloodied youths turning toward the flash.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The struck captive now lies motionless amid fresh blood in the slaughterhouse. Jihu still wears her antler-bone pendant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 청년 (한국인 20~30대 남성, 짧은 검은 머리, 그을린 얼굴, 거친 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 9,
   "A": 4,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 9,
    "verdict_ko": "실내에서 창문을 향하는 카메라 위치를 정확히 지켰으며, 창문 쪽 플래시 불빛을 향해 일제히 고개를 돌린 청년들과 쓰러진 포로의 상태까지 완벽하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "카메라가 실내에서 창문을 향해야 한다는 위치 지침을 어기고 창밖에서 실내를 들여다보는 구도로 연출된 하드 위반(Hard Violation)이 있습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "카메라가 창문이 아닌 뒷벽을 향하고 있으며, 인물들도 창문 쪽이 아닌 카메라를 바라보고 있어 행동 지문 및 카메라 방향 지침을 모두 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S32sh10_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 청년: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:756328>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "청년들의 얼굴과 옷에 피가 약간만 묻어 있어, 프롬프트가 요구한 '피범벅' 상태가 제대로 표현되지 않음.",
     "fix_en": "Add heavy red blood splatters and thick smears all over the faces and clothing of the three young men.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "왼쪽 배경에 서 있는 남자의 복부 앞쪽 허공에 식칼 모양의 물체가 부자연스럽게 떠서 겹쳐 있음.",
     "fix_en": "Remove the floating meat cleaver shape overlapping the background man's torso and the table.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add heavy red blood splatters and thick smears all over the faces and clothing of the three young men.\n- Remove the floating meat cleaver shape overlapping the background man's torso and the table.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S32sh10"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S33sh6::bgfirst_bg": {
  "input_fingerprint": "304bd0ebbb8c0a29",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 오솔길에 다급히 멈춰 선 수리영과, 맞은편 숲 경계에 우뚝 서서 그녀를 마주 보고 있는 산들의 전경.\n\nLOCATION (lock): On a small open trail where dense forest gives way at two opposing edges.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From eye level beside the path, frame 김지후(강수리영) on the left and 고산들 on the right in a static wide profile composition, with the open path separating them and the opposing forest boundaries holding each figure at an edge. 김지후(강수리영) is caught just after stopping, pitched forward from her run, while 고산들 remains planted across from her; the perpendicular view makes the distance between them the dominant visual fact.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, midground, looks toward 고산들 across the path; 고산들 in the middle-right of the frame, midground, looks toward 김지후(강수리영) across the path; open path between them in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 오솔길 (open between the opposing forest edges); used as Forms the central strip of negative space that measures the separation between the two figures; 맞은편 숲 경계 (김지후(강수리영)와 고산들이 서로 다른 쪽에서 마주한 상태); used as Frames each character from opposite sides and establishes where they have emerged.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 오솔길에 다급히 멈춰 선 수리영과, 맞은편 숲 경계에 우뚝 서서 그녀를 마주 보고 있는 산들의 전경.\n\nLOCATION (lock): On a small open trail where dense forest gives way at two opposing edges.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From eye level beside the path, frame 김지후(강수리영) on the left and 고산들 on the right in a static wide profile composition, with the open path separating them and the opposing forest boundaries holding each figure at an edge. 김지후(강수리영) is caught just after stopping, pitched forward from her run, while 고산들 remains planted across from her; the perpendicular view makes the distance between them the dominant visual fact.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, midground, looks toward 고산들 across the path; 고산들 in the middle-right of the frame, midground, looks toward 김지후(강수리영) across the path; open path between them in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 오솔길 (open between the opposing forest edges); used as Forms the central strip of negative space that measures the separation between the two figures; 맞은편 숲 경계 (김지후(강수리영)와 고산들이 서로 다른 쪽에서 마주한 상태); used as Frames each character from opposite sides and establishes where they have emerged.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S33sh6__bgfirst_bg.png",
  "asset_id": "c02068ac-78ae-404e-920b-866f1ca11ed5",
  "input_asset_ids": [
   "9a19cba3-6342-4807-b242-bda1663ad31d",
   "f055d50e-ca61-488c-8364-ff8e46aea3f7"
  ]
 },
 "S33sh6": {
  "input_fingerprint": "165387cffeb8239e",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 오솔길에 다급히 멈춰 선 수리영과, 맞은편 숲 경계에 우뚝 서서 그녀를 마주 보고 있는 산들의 전경.\n\nLOCATION (lock): On a small open trail where dense forest gives way at two opposing edges. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang remains fallen below the slope after disappearing over the edge. Jihu’s antler-bone pendant is no longer securely accounted for after this night.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 오솔길에 다급히 멈춰 선 수리영과, 맞은편 숲 경계에 우뚝 서서 그녀를 마주 보고 있는 산들의 전경.\n\nLOCATION (lock): On a small open trail where dense forest gives way at two opposing edges. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang remains fallen below the slope after disappearing over the edge. Jihu’s antler-bone pendant is no longer securely accounted for after this night.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 오솔길에 다급히 멈춰 선 수리영과, 맞은편 숲 경계에 우뚝 서서 그녀를 마주 보고 있는 산들의 전경.\n\nLOCATION (lock): On a small open trail where dense forest gives way at two opposing edges. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang remains fallen below the slope after disappearing over the edge. Jihu’s antler-bone pendant is no longer securely accounted for after this night.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S33sh6__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S33sh6.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L26B02.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "다급히 멈춰 선 수리영의 역동적인 포즈를 잘 구현했으며, 레퍼런스의 의상과 뿔 목걸이를 정확히 유지해 훌륭한 일치도를 보입니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "인물의 포즈가 다소 정적이어서 다급한 느낌이 부족하며, 레퍼런스와 다른 형태의 목걸이가 묘사되어 복장 일치도가 떨어집니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "다급히 멈춰 선 수리영의 역동적인 포즈를 잘 구현했으며, 레퍼런스의 의상과 뿔 목걸이를 정확히 유지해 훌륭한 일치도를 보입니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "인물의 포즈가 다소 정적이어서 다급한 느낌이 부족하며, 레퍼런스와 다른 형태의 목걸이가 묘사되어 복장 일치도가 떨어집니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "다급히 멈춰 선 수리영의 역동적인 동작과 맞은편 인물과의 대치 상황을 샷 텍스트에 맞게 훌륭한 전경(wide shot)으로 연출했습니다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "수리영이 다급하게 멈춰 섰다는 지문과 달리 단순히 서 있는 듯한 자세로 묘사되어 극적인 연출이 부족합니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "다급히 멈춰 선 수리영의 역동적인 동작과 맞은편 인물과의 대치 상황을 샷 텍스트에 맞게 훌륭한 전경(wide shot)으로 연출했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "수리영이 다급하게 멈춰 섰다는 지문과 달리 단순히 서 있는 듯한 자세로 묘사되어 극적인 연출이 부족합니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 10
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 10
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "다급히 멈춰 선 수리영의 역동적인 포즈를 잘 구현했으며, 레퍼런스의 의상과 뿔 목걸이를 정확히 유지해 훌륭한 일치도를 보입니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "인물의 포즈가 다소 정적이어서 다급한 느낌이 부족하며, 레퍼런스와 다른 형태의 목걸이가 묘사되어 복장 일치도가 떨어집니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L26B02.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "허용된 인물 목록(PEOPLE)에 없는 남자가 오른쪽에 렌더링되었습니다 ('never anyone else' 규칙 위반).",
     "fix_en": "Remove the man on the right entirely, restoring the forest background to match the empty reference image.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "휴대 상태(Carried State) 지침에 뿔 펜던트가 없어야 한다고 명시되어 있으나, 여성의 목에 펜던트가 그려져 있습니다.",
     "fix_en": "Remove the antler-bone pendant from the woman's neck, leaving only the dark sweater.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the man on the right entirely, restoring the forest background to match the empty reference image.\n- Remove the antler-bone pendant from the woman's neck, leaving only the dark sweater.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S33sh6__bgfirst_bg.png",
   "bg_asset_id": "c02068ac-78ae-404e-920b-866f1ca11ed5",
   "bg_record_key": "S33sh6::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S33sh12": {
  "input_fingerprint": "8f09c8d045afd638",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 섬뜩한 황금색의 직사각형 형태로 뚜렷하게 고정된 산들의 눈동자 클로즈업.\n\nLOCATION (lock): On the open forest trail where the pursuers converge from the surrounding trees. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the moonlit forest-path look and the man's position facing the woman. Exclude the broad path composition and all pursuers from the eye close-up.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang remains below the slope, and Sandeul’s eyes have transformed to golden rectangular pupils.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 5,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "황금색 직사각형 동공과 양안 클로즈업 프레이밍을 지시문대로 가장 정확하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "직사각형 동공은 표현했으나, 요구된 클로즈업 샷을 단일 눈 익스트림 클로즈업으로 과도하게 좁혀 렌더링했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "눈 주위에 허구의 직사각형 그래픽(UI 마커)을 추가한 것은 심각한 규정 위반(Hard Violation)입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S33sh6_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1225321>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트와 이전 샷 상태에서 '황금색의 직사각형 눈동자(동공)'를 요구했으나, 이미지에서는 홍채만 황금색이고 동공은 검은색 직사각형으로 표현되었습니다.",
     "fix_en": "Change the color of the black rectangular pupils in the center of the eyes to golden.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the color of the black rectangular pupils in the center of the eyes to golden.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S33sh6"
  }
 },
 "S33sh13": {
  "input_fingerprint": "5c8d2496f8ba274c",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 목을 움켜쥔 채 수리영을 잡아먹을 듯 무시무시하게 노려보는 산들의 눈빛.\n\nLOCATION (lock): On the same narrow forest trail at the edge of the dense undergrowth. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the forest edge, moonlit path, and continuing confrontation positions. Exclude the earlier separated stance and show the required neck-grab action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang remains below the slope; Jihu becomes unconscious and limp in Sandeul’s grip.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 4,
   "B": 3,
   "C": 7
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "목을 조르는 액션과 산들의 매서운 눈빛 등 샷 텍스트의 핵심 요구사항이 완전히 누락되었습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "인물들 사이에 복제된 여성의 얼굴이 존재하는 치명적인 추가 신체 오류(Hard Violation)가 발생했습니다."
   },
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "축 늘어진 지후의 상태와 목을 움켜쥔 산들의 강렬한 눈빛 등 지시된 액션과 상황이 가장 충실하게 연출되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S33sh12_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "산들이 수리영을 노려보지 않고 카메라 정면을 응시하고 있습니다.",
     "fix_en": "Redirect the man's eyes downwards so he is glaring fiercely directly at the woman's face.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "지후(수리영)가 캐릭터 레퍼런스에 지정된 짙은 색 스웨터가 아닌 밝은 색 전통 의상을 입고 있습니다.",
     "fix_en": "Change the woman's clothing to a dark green cable-knit sweater to match her character reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redirect the man's eyes downwards so he is glaring fiercely directly at the woman's face.\n- Change the woman's clothing to a dark green cable-knit sweater to match her character reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S33sh12"
  }
 },
 "groupbg::승용차_실내": {
  "input_fingerprint": "c422f42fd313f51b",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "승용차_실내",
    "tags": [
     "S34sh1",
     "S34sh2"
    ]
   },
   "context_sig": "1a4b630684d63534"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: In the driver's seat of a compact car speeding along a daytime rural highway.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 해송 숲길 및 숲속: 해송이 울창하게 우거져 빛이 잘 들지 않는 빽빽하고 좁은 흙길과 숲의 심층부. (특징: 빽빽한 해송 나무 기둥들; 무성하고 어두운 수풀; 바닥에 깔린 흙과 솔잎; ‘수리영’이 적힌 이끼 낀 비석들; 비석 앞에 놓인 썩은 핏빛 육고기 덩어리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실내. 달리는 아반테 - 낮\n- 운전대를 잡고 있는 인우는 걱정 가득한 얼굴이다.\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: In the driver's seat of a compact car speeding along a daytime rural highway.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 해송 숲길 및 숲속: 해송이 울창하게 우거져 빛이 잘 들지 않는 빽빽하고 좁은 흙길과 숲의 심층부. (특징: 빽빽한 해송 나무 기둥들; 무성하고 어두운 수풀; 바닥에 깔린 흙과 솔잎; ‘수리영’이 적힌 이끼 낀 비석들; 비석 앞에 놓인 썩은 핏빛 육고기 덩어리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실내. 달리는 아반테 - 낮\n- 운전대를 잡고 있는 인우는 걱정 가득한 얼굴이다.\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_승용차_실내_297d63.png",
  "asset_id": "5a7c69a4-18e4-4f4d-816d-f564b846f053",
  "input_asset_ids": [
   "b0e27061-878f-4ca2-9646-2caf424aecde"
  ],
  "origin_tag": "S34sh1",
  "place_text": "In the driver's seat of a compact car speeding along a daytime rural highway.",
  "origin_inputs": {
   "place_text": "In the driver's seat of a compact car speeding along a daytime rural highway.",
   "time_of_day_en": "day",
   "conti_asset_id": "b0e27061-878f-4ca2-9646-2caf424aecde"
  }
 },
 "S34sh1::bgfirst_bg": {
  "input_fingerprint": "cb1872b5072ce198",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 한쪽 팔에 깁스를 한 인우가 운전석에서 미간을 잔뜩 찌푸린 채 전방을 주시하는 측면.\n\nLOCATION (lock): In the driver's seat of a compact car speeding along a daytime rural highway.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the front passenger area at 정인우’s eye level, hold a medium profile composition with his face in the upper-left field, the steering wheel and cast arm crossing the lower center, and the road visible through the windshield beyond. Keep the camera static and allow his narrowed brow and fixed forward gaze to carry the tension rather than altering the driving axis.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 앞유리 (차량이 지방 국도를 달리는 중) — The road ahead is visible through the windshield from the passenger-side camera position; used as Carries the forward sightline beyond 정인우’s profile and confirms that the car is traveling on the road; 운전대 (정인우가 잡고 있는 상태) — Its driver-facing side is turned toward 정인우, with the rim seen obliquely by the passenger-side camera; used as Anchors the cast arm in the lower portion of the composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 한쪽 팔에 깁스를 한 인우가 운전석에서 미간을 잔뜩 찌푸린 채 전방을 주시하는 측면.\n\nLOCATION (lock): In the driver's seat of a compact car speeding along a daytime rural highway.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the front passenger area at 정인우’s eye level, hold a medium profile composition with his face in the upper-left field, the steering wheel and cast arm crossing the lower center, and the road visible through the windshield beyond. Keep the camera static and allow his narrowed brow and fixed forward gaze to carry the tension rather than altering the driving axis.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 앞유리 (차량이 지방 국도를 달리는 중) — The road ahead is visible through the windshield from the passenger-side camera position; used as Carries the forward sightline beyond 정인우’s profile and confirms that the car is traveling on the road; 운전대 (정인우가 잡고 있는 상태) — Its driver-facing side is turned toward 정인우, with the rim seen obliquely by the passenger-side camera; used as Anchors the cast arm in the lower portion of the composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S34sh1__bgfirst_bg.png",
  "asset_id": "43f13a4b-28aa-4780-baaf-b93fb34bade6",
  "input_asset_ids": [
   "b0e27061-878f-4ca2-9646-2caf424aecde",
   "5a7c69a4-18e4-4f4d-816d-f564b846f053"
  ]
 },
 "S34sh1": {
  "input_fingerprint": "905c212828a6d51d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 한쪽 팔에 깁스를 한 인우가 운전석에서 미간을 잔뜩 찌푸린 채 전방을 주시하는 측면.\n\nLOCATION (lock): In the driver's seat of a compact car speeding along a daytime rural highway. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s previously injured arm is now immobilized in a cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 한쪽 팔에 깁스를 한 인우가 운전석에서 미간을 잔뜩 찌푸린 채 전방을 주시하는 측면.\n\nLOCATION (lock): In the driver's seat of a compact car speeding along a daytime rural highway. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s previously injured arm is now immobilized in a cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 한쪽 팔에 깁스를 한 인우가 운전석에서 미간을 잔뜩 찌푸린 채 전방을 주시하는 측면.\n\nLOCATION (lock): In the driver's seat of a compact car speeding along a daytime rural highway. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s previously injured arm is now immobilized in a cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S34sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S34sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_승용차_실내_297d63.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "레퍼런스의 차량 내부 구조(대시보드 등)를 매우 정확하게 반영했으며, 깁스를 한 상태로 미간을 찌푸린 측면 모습 등 프롬프트 지시를 충실히 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "미간을 찌푸린 표정은 잘 연출되었으나, 레퍼런스의 차량 내부 디테일을 놓쳤고 빈 슬링(목걸이)이 추가되었으며 팔을 교차해 핸들을 잡는 다소 부자연스러운 자세가 아쉽습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "레퍼런스의 차량 내부 구조(대시보드 등)를 매우 정확하게 반영했으며, 깁스를 한 상태로 미간을 찌푸린 측면 모습 등 프롬프트 지시를 충실히 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "미간을 찌푸린 표정은 잘 연출되었으나, 레퍼런스의 차량 내부 디테일을 놓쳤고 빈 슬링(목걸이)이 추가되었으며 팔을 교차해 핸들을 잡는 다소 부자연스러운 자세가 아쉽습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "해부학적으로 불가능하게 팔이 가슴을 관통하고 손의 구조가 왜곡되어 하드 위반입니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "요청된 측면 구도, 찌푸린 미간, 한쪽 팔의 깁스 등 모든 지시사항을 물리적 오류 없이 정확하게 구현했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "해부학적으로 불가능하게 팔이 가슴을 관통하고 손의 구조가 왜곡되어 하드 위반입니다."
     },
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "요청된 측면 구도, 찌푸린 미간, 한쪽 팔의 깁스 등 모든 지시사항을 물리적 오류 없이 정확하게 구현했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 7
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 7
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "레퍼런스의 차량 내부 구조(대시보드 등)를 매우 정확하게 반영했으며, 깁스를 한 상태로 미간을 찌푸린 측면 모습 등 프롬프트 지시를 충실히 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "미간을 찌푸린 표정은 잘 연출되었으나, 레퍼런스의 차량 내부 디테일을 놓쳤고 빈 슬링(목걸이)이 추가되었으며 팔을 교차해 핸들을 잡는 다소 부자연스러운 자세가 아쉽습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_승용차_실내_297d63.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트에 '미간을 잔뜩 찌푸린 채'라고 명시되어 있으나, 캐릭터의 미간이 평온하고 찌푸린 표정이 잘 드러나지 않습니다.",
     "fix_en": "Furrow the character's eyebrows deeply to create a strong, clear frown.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Furrow the character's eyebrows deeply to create a strong, clear frown.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S34sh1__bgfirst_bg.png",
   "bg_asset_id": "43f13a4b-28aa-4780-baaf-b93fb34bade6",
   "bg_record_key": "S34sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "승용차_실내",
   "groupbg_asset_id": "5a7c69a4-18e4-4f4d-816d-f564b846f053"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S34sh2": {
  "input_fingerprint": "5bcf73113c3c13db",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 인우의 자동차 핸드폰 거치대 화면에 십여 개의 '김지후' 발신 실패 기록이 빽빽하게 뜬 찰나.\n\nLOCATION (lock): Inside the moving compact car's front cabin, at the mounted phone beside the driver. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo continues to wear the cast on his injured arm.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 5,
   "B": 7,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "운전자의 깁스와 주행 연속성은 유지했으나, 화면에 '김지후'가 아닌 무작위 이름들이 섞여 있어 텍스트 지시를 빗나갔습니다."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "화면에 십여 개의 '김지후' 발신 실패 기록을 가장 정확히 반복 출력했으며, 이전 샷의 깁스 연속성도 훌륭히 유지했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "화면의 이름은 일치하나, 주행 중인 차량의 운전대에서 운전자가 완전히 사라지는 치명적인 물리적/연속성 오류가 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S34sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지문이 인물이나 신체 부위를 명시하지 않은 한 프레임에 나타나지 않아야 한다는 규칙이 있으나, 운전대를 잡고 있는 손과 팔이 보임.",
     "fix_en": "Remove the hand and arm holding the steering wheel entirely, leaving only the bare steering wheel and dashboard visible.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the hand and arm holding the steering wheel entirely, leaving only the bare steering wheel and dashboard visible.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev만 (배경 전용·공유 계획)",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S34sh1"
  },
  "lane_policy": "share_plan_prev_bgonly"
 },
 "groupbg::꿈속_숲길": {
  "input_fingerprint": "7ade28d24c54ca10",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "꿈속_숲길",
    "tags": [
     "S35sh3",
     "S35sh6"
    ]
   },
   "context_sig": "aed10925690b8900"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On a dreamlike daytime woodland trail, with the daughter and mother separated along the same path.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n아반테 내부: 평범한 현대식 준중형 승용차의 내부 공간. (특징: 직물/가죽 차량 시트; 운전대; 내비게이션 앱 지도가 켜진 스마트폰 거치대)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 숲속 오솔길 - 낮(수리영의 꿈)\n- 수리영이 오솔길에 홀로 서 있다.\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On a dreamlike daytime woodland trail, with the daughter and mother separated along the same path.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n아반테 내부: 평범한 현대식 준중형 승용차의 내부 공간. (특징: 직물/가죽 차량 시트; 운전대; 내비게이션 앱 지도가 켜진 스마트폰 거치대)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 숲속 오솔길 - 낮(수리영의 꿈)\n- 수리영이 오솔길에 홀로 서 있다.\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_꿈속_숲길_87ae72.png",
  "asset_id": "027584ab-9536-42a3-9ea6-7a2d18ba5a8c",
  "input_asset_ids": [
   "dfdbc4d8-e360-442a-8900-5d32405de73f"
  ],
  "origin_tag": "S35sh3",
  "place_text": "On a dreamlike daytime woodland trail, with the daughter and mother separated along the same path.",
  "origin_inputs": {
   "place_text": "On a dreamlike daytime woodland trail, with the daughter and mother separated along the same path.",
   "time_of_day_en": "day",
   "conti_asset_id": "dfdbc4d8-e360-442a-8900-5d32405de73f"
  }
 },
 "S35sh3::bgfirst_bg": {
  "input_fingerprint": "8a2a2b8028ae72dc",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 수리영이 간절한 표정으로 민숙을 향해 한 팔을 뻗은 채, 한쪽 발이 허공에 떠 있는 달리기 중간 자세.\n\nLOCATION (lock): On a dreamlike daytime woodland trail, with the daughter and mother separated along the same path.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track beside 김지후(강수리영) at eye level into a medium-full three-quarter frontal view, catching one foot suspended, her torso pitched forward, and one arm reaching toward 강민숙(홍은미) along the path. 김지후(강수리영) occupies the near left-center while 강민숙(홍은미) remains small and distant on the upper path, preserving the dream’s unchanged gap despite the running motion.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, foreground, moves toward 강민숙(홍은미) on the upper path; 강민숙(홍은미) in the upper-center of the frame, background, looks toward 김지후(강수리영) running below; unchanged stretch of path in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 금월도 숲속 오솔길 (두 사람 사이의 거리가 좁혀지지 않는 꿈속 상태); used as Creates the visible but supernaturally unchanging distance between daughter and mother; 위쪽 길 (강민숙(홍은미)이 멀찍이 있는 상태); used as Defines the distant position where 강민숙(홍은미) remains fixed.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 수리영이 간절한 표정으로 민숙을 향해 한 팔을 뻗은 채, 한쪽 발이 허공에 떠 있는 달리기 중간 자세.\n\nLOCATION (lock): On a dreamlike daytime woodland trail, with the daughter and mother separated along the same path.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track beside 김지후(강수리영) at eye level into a medium-full three-quarter frontal view, catching one foot suspended, her torso pitched forward, and one arm reaching toward 강민숙(홍은미) along the path. 김지후(강수리영) occupies the near left-center while 강민숙(홍은미) remains small and distant on the upper path, preserving the dream’s unchanged gap despite the running motion.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, foreground, moves toward 강민숙(홍은미) on the upper path; 강민숙(홍은미) in the upper-center of the frame, background, looks toward 김지후(강수리영) running below; unchanged stretch of path in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 금월도 숲속 오솔길 (두 사람 사이의 거리가 좁혀지지 않는 꿈속 상태); used as Creates the visible but supernaturally unchanging distance between daughter and mother; 위쪽 길 (강민숙(홍은미)이 멀찍이 있는 상태); used as Defines the distant position where 강민숙(홍은미) remains fixed.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S35sh3__bgfirst_bg.png",
  "asset_id": "a445a522-6ae0-401b-9ef8-d7313f12f068",
  "input_asset_ids": [
   "dfdbc4d8-e360-442a-8900-5d32405de73f",
   "027584ab-9536-42a3-9ea6-7a2d18ba5a8c"
  ]
 },
 "S35sh3": {
  "input_fingerprint": "592bf209a5566a41",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 수리영이 간절한 표정으로 민숙을 향해 한 팔을 뻗은 채, 한쪽 발이 허공에 떠 있는 달리기 중간 자세.\n\nLOCATION (lock): On a dreamlike daytime woodland trail, with the daughter and mother separated along the same path. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 수리영이 간절한 표정으로 민숙을 향해 한 팔을 뻗은 채, 한쪽 발이 허공에 떠 있는 달리기 중간 자세.\n\nLOCATION (lock): On a dreamlike daytime woodland trail, with the daughter and mother separated along the same path. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 수리영이 간절한 표정으로 민숙을 향해 한 팔을 뻗은 채, 한쪽 발이 허공에 떠 있는 달리기 중간 자세.\n\nLOCATION (lock): On a dreamlike daytime woodland trail, with the daughter and mother separated along the same path. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S35sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S35sh3.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_꿈속_숲길_87ae72.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "민숙을 향해 팔을 뻗으며 한 발을 떼고 달리는 자세의 방향성과 공간적 관계를 지문 그대로 정확히 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "인물이 뒤에 있는 민숙이 아닌 카메라를 향해 팔을 뻗고 달려 '민숙을 향해'라는 핵심 행동 방향을 어겼습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "민숙을 향해 팔을 뻗으며 한 발을 떼고 달리는 자세의 방향성과 공간적 관계를 지문 그대로 정확히 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "인물이 뒤에 있는 민숙이 아닌 카메라를 향해 팔을 뻗고 달려 '민숙을 향해'라는 핵심 행동 방향을 어겼습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "수리영이 민숙을 향하지 않고 등진 채 반대 방향으로 달리고 있어 '민숙을 향해 팔을 뻗은'이라는 핵심 지시를 위반했습니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "한쪽 발이 허공에 뜬 달리기 자세로 멀리 있는 민숙을 향해 팔을 뻗는 동선과 동작을 프롬프트에 맞게 정확히 구현했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "수리영이 민숙을 향하지 않고 등진 채 반대 방향으로 달리고 있어 '민숙을 향해 팔을 뻗은'이라는 핵심 지시를 위반했습니다."
     },
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "한쪽 발이 허공에 뜬 달리기 자세로 멀리 있는 민숙을 향해 팔을 뻗는 동선과 동작을 프롬프트에 맞게 정확히 구현했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 8
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 15,
   "B": 8
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "민숙을 향해 팔을 뻗으며 한 발을 떼고 달리는 자세의 방향성과 공간적 관계를 지문 그대로 정확히 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "인물이 뒤에 있는 민숙이 아닌 카메라를 향해 팔을 뻗고 달려 '민숙을 향해'라는 핵심 행동 방향을 어겼습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_꿈속_숲길_87ae72.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "달리는 인물의 다리 교차 방향이 스케치와 반대로 되어 있으며(왼다리가 앞, 오른다리가 뒤여야 함), '한쪽 발이 허공에 떠 있는'이라는 지문과 달리 두 발 모두 공중에 떠 있습니다.",
     "fix_en": "Redraw the runner's lower body to match the sketch: extend the left leg (closest to the camera) forward with its foot planted firmly on the ground, and bend the right leg backward in the air.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the runner's lower body to match the sketch: extend the left leg (closest to the camera) forward with its foot planted firmly on the ground, and bend the right leg backward in the air.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S35sh3__bgfirst_bg.png",
   "bg_asset_id": "a445a522-6ae0-401b-9ef8-d7313f12f068",
   "bg_record_key": "S35sh3::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "꿈속_숲길",
   "groupbg_asset_id": "027584ab-9536-42a3-9ea6-7a2d18ba5a8c"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S35sh6": {
  "input_fingerprint": "adbb3647dee0e2e8",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 민숙의 눈동자가 섬뜩한 황금색 직사각형으로 변한 찰나의 클로즈업.\n\nLOCATION (lock): On the upper stretch of the dreamlike forest trail in daylight. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the bright dreamlike forest path, fixed distance between mother and daughter, and unreal stillness. Exclude the daughter's running pose and isolate the mother's transformed eyes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "정상적인 인간의 눈 구조를 유지하라는 지침을 엄격히 준수하여, 동공을 물리적으로 변형하는 대신 황금색 직사각형을 눈에 비친 반사광으로 현실감 있게 연출했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "정상적인 눈 해부학 구조 유지 지침을 위반했으며, '황금색 직사각형'이라는 명확한 묘사도 검은색 수직 동공으로 잘못 표현했습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "눈을 고립시키는 클로즈업 샷은 훌륭하나, 동공 자체가 발광하는 황금색 사각형으로 변형되어 정상적인 눈 구조를 유지하라는 지침을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S35sh3_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "눈동자 안에 말 그대로 빛나는 황금색 직사각형이 묘사되어, 눈을 비유적 표현이 아닌 해부학적으로 변형시키지 말라는 지시(정상적인 홍채와 동공 유지)를 위반했습니다.",
     "fix_en": "Restore the eyes to anatomically normal human eyes with natural round black pupils and regular irises, removing the glowing golden rectangular shapes.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore the eyes to anatomically normal human eyes with natural round black pupils and regular irises, removing the glowing golden rectangular shapes.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S35sh3"
  }
 },
 "groupbg::식당_뒷방": {
  "input_fingerprint": "76f834c45b9c65a2",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "식당_뒷방",
    "tags": [
     "S36sh1",
     "S36sh4",
     "S36sh5",
     "S50sh2",
     "S50sh3"
    ]
   },
   "context_sig": "19388e654f7952fa"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On neatly arranged bedding in the restaurant's rear bedroom, warmed by daylight entering through the window.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n꿈속 금월도 숲속 오솔길: 거리감과 공간감이 왜곡되어 아무리 걸어도 제자리인 듯한 비현실적인 숲속 오솔길. (특징: 거리가 왜곡된 양옆의 나무와 수풀; 길어 보이는 좁은 오솔길) / 해안마을 횟집 야외 좌석: 밤바다가 내다보이는 육지 해안가 횟집의 야외 테이블 공간. (특징: 야외 테이블과 의자; 테이블 위 소주병과 술잔; 횟집 간판 불빛)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실내. 반월 식당 뒷방 - 낮\n- 창을 타고 따스한 햇살이 스며드는 정갈하게 정돈된 방이다.\n- 실내. 반월 식당 뒷방 - 밤\n\nTIME OF DAY (lock): day, warm sunlight.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On neatly arranged bedding in the restaurant's rear bedroom, warmed by daylight entering through the window.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n꿈속 금월도 숲속 오솔길: 거리감과 공간감이 왜곡되어 아무리 걸어도 제자리인 듯한 비현실적인 숲속 오솔길. (특징: 거리가 왜곡된 양옆의 나무와 수풀; 길어 보이는 좁은 오솔길) / 해안마을 횟집 야외 좌석: 밤바다가 내다보이는 육지 해안가 횟집의 야외 테이블 공간. (특징: 야외 테이블과 의자; 테이블 위 소주병과 술잔; 횟집 간판 불빛)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실내. 반월 식당 뒷방 - 낮\n- 창을 타고 따스한 햇살이 스며드는 정갈하게 정돈된 방이다.\n- 실내. 반월 식당 뒷방 - 밤\n\nTIME OF DAY (lock): day, warm sunlight.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_식당_뒷방_dd80df.png",
  "asset_id": "02ca0ffe-5d7b-4fd6-8e7e-c0f453411750",
  "input_asset_ids": [
   "f2804adb-08bb-41e4-acfe-ddccdc984060"
  ],
  "origin_tag": "S36sh1",
  "place_text": "On neatly arranged bedding in the restaurant's rear bedroom, warmed by daylight entering through the window.",
  "origin_inputs": {
   "place_text": "On neatly arranged bedding in the restaurant's rear bedroom, warmed by daylight entering through the window.",
   "time_of_day_en": "day, warm sunlight",
   "conti_asset_id": "f2804adb-08bb-41e4-acfe-ddccdc984060"
  }
 },
 "S36sh1::bgfirst_bg": {
  "input_fingerprint": "3723766180faac69",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 정갈한 뒷방 이불 위에 누운 수리영이 가위에 눌려 눈만 번쩍 뜬 채 몸이 굳어있는 찰나.\n\nLOCATION (lock): On neatly arranged bedding in the restaurant's rear bedroom, warmed by daylight entering through the window.\n\nTIME OF DAY (lock): day, warm sunlight.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At bedding height beside 김지후(강수리영), hold the opening position nearly still in a close view angled across her rigid upper body toward her suddenly opened eyes. Her face occupies the upper-right portion and her immobilized torso extends diagonally through the lower frame, with enough bedding visible to establish that she has awakened lying down.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 김지후(강수리영) in the middle-right of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 정갈한 이불 (정돈된 뒷방에 펼쳐진 상태); used as Supports the low camera height and outlines the rigid body without distracting from the opened eyes; 창 (햇살이 스며드는 상태) — Warm sunlight enters the room through the window; the exterior beyond is not emphasized; used as Appears as limited room context and motivates the established daytime illumination.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 정갈한 뒷방 이불 위에 누운 수리영이 가위에 눌려 눈만 번쩍 뜬 채 몸이 굳어있는 찰나.\n\nLOCATION (lock): On neatly arranged bedding in the restaurant's rear bedroom, warmed by daylight entering through the window.\n\nTIME OF DAY (lock): day, warm sunlight.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At bedding height beside 김지후(강수리영), hold the opening position nearly still in a close view angled across her rigid upper body toward her suddenly opened eyes. Her face occupies the upper-right portion and her immobilized torso extends diagonally through the lower frame, with enough bedding visible to establish that she has awakened lying down.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 김지후(강수리영) in the middle-right of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 정갈한 이불 (정돈된 뒷방에 펼쳐진 상태); used as Supports the low camera height and outlines the rigid body without distracting from the opened eyes; 창 (햇살이 스며드는 상태) — Warm sunlight enters the room through the window; the exterior beyond is not emphasized; used as Appears as limited room context and motivates the established daytime illumination.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S36sh1__bgfirst_bg.png",
  "asset_id": "b3b27c94-5999-43ad-af1b-18450a9ca9ad",
  "input_asset_ids": [
   "f2804adb-08bb-41e4-acfe-ddccdc984060",
   "02ca0ffe-5d7b-4fd6-8e7e-c0f453411750"
  ]
 },
 "S36sh1": {
  "input_fingerprint": "7781d1a85cd04930",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, warm sunlight.\n\nSHOT TEXT (authoritative, Korean): 정갈한 뒷방 이불 위에 누운 수리영이 가위에 눌려 눈만 번쩍 뜬 채 몸이 굳어있는 찰나.\n\nLOCATION (lock): On neatly arranged bedding in the restaurant's rear bedroom, warmed by daylight entering through the window. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is stretched out on the neatly made bed with her head supported by the bedding and all four limbs held rigid and motionless by sleep paralysis.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant is missing, and she remains stretched out on the neatly made bed with her head supported by the bedding and all four limbs held rigid and motionless by sleep paralysis, where she was placed after losing consciousness.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, warm sunlight.\n\nSHOT TEXT (authoritative, Korean): 정갈한 뒷방 이불 위에 누운 수리영이 가위에 눌려 눈만 번쩍 뜬 채 몸이 굳어있는 찰나.\n\nLOCATION (lock): On neatly arranged bedding in the restaurant's rear bedroom, warmed by daylight entering through the window. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is stretched out on the neatly made bed with her head supported by the bedding and all four limbs held rigid and motionless by sleep paralysis.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant is missing, and she remains stretched out on the neatly made bed with her head supported by the bedding and all four limbs held rigid and motionless by sleep paralysis, where she was placed after losing consciousness.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, warm sunlight.\n\nSHOT TEXT (authoritative, Korean): 정갈한 뒷방 이불 위에 누운 수리영이 가위에 눌려 눈만 번쩍 뜬 채 몸이 굳어있는 찰나.\n\nLOCATION (lock): On neatly arranged bedding in the restaurant's rear bedroom, warmed by daylight entering through the window. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is stretched out on the neatly made bed with her head supported by the bedding and all four limbs held rigid and motionless by sleep paralysis.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant is missing, and she remains stretched out on the neatly made bed with her head supported by the bedding and all four limbs held rigid and motionless by sleep paralysis, where she was placed after losing consciousness.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S36sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S36sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_식당_뒷방_dd80df.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 장소에서 가위에 눌려 눈만 뜬 채 굳어있는 인물의 자세와 표정을 정확히 연출했으며, 목걸이가 없는 상태도 잘 반영했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "하나의 몸에 두 개의 머리가 생성되는 치명적인 신체 훼손 및 구조적 오류(Hard Violation)가 발생하여 실격입니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 장소에서 가위에 눌려 눈만 뜬 채 굳어있는 인물의 자세와 표정을 정확히 연출했으며, 목걸이가 없는 상태도 잘 반영했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "하나의 몸에 두 개의 머리가 생성되는 치명적인 신체 훼손 및 구조적 오류(Hard Violation)가 발생하여 실격입니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "신체 양끝에 머리가 두 개 달린 기형적인 해부학적 오류가 발생하여 최우선 하드 위반입니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 장소에서 가위에 눌려 눈만 번쩍 뜬 채 굳어있는 인물의 자세와 표정을 매우 충실하게 구현했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 2,
      "verdict_ko": "신체 양끝에 머리가 두 개 달린 기형적인 해부학적 오류가 발생하여 최우선 하드 위반입니다."
     },
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 장소에서 가위에 눌려 눈만 번쩍 뜬 채 굳어있는 인물의 자세와 표정을 매우 충실하게 구현했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 5
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지정된 장소에서 가위에 눌려 눈만 뜬 채 굳어있는 인물의 자세와 표정을 정확히 연출했으며, 목걸이가 없는 상태도 잘 반영했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "하나의 몸에 두 개의 머리가 생성되는 치명적인 신체 훼손 및 구조적 오류(Hard Violation)가 발생하여 실격입니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_식당_뒷방_dd80df.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "수면 마비로 몸이 굳어 중력에 따라 늘어져야 한다는 지침과 달리, 목에 힘을 주어 머리를 베개에서 앞으로 들어 올리고 있습니다.",
     "fix_en": "Drop her head backward so her neck relaxes completely and the back of her head rests flat and limply against the pillow.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Drop her head backward so her neck relaxes completely and the back of her head rests flat and limply against the pillow.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S36sh1__bgfirst_bg.png",
   "bg_asset_id": "b3b27c94-5999-43ad-af1b-18450a9ca9ad",
   "bg_record_key": "S36sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "식당_뒷방",
   "groupbg_asset_id": "02ca0ffe-5d7b-4fd6-8e7e-c0f453411750"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S36sh4": {
  "input_fingerprint": "a71ea450c9bf7118",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, warm sunlight.\n\nSHOT TEXT (authoritative, Korean): 다급한 표정으로 자신의 텅 빈 목덜미에 손을 얹은 채 멈칫한 수리영의 찰나.\n\nLOCATION (lock): Inside the sunlit, orderly rear bedroom of the restaurant. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the tidy sunlit room, bedding, warm window light, and the woman's lying position. Exclude the sleep-paralysis stiffness and emphasize her hand checking the empty neck.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s neck remains bare; the antler-bone pendant is missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 6,
   "B": 9,
   "C": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "목덜미를 확인하는 다급한 표정은 잘 표현되었으나, 레퍼런스에서 유지하도록 지시된 '누워있는 자세'를 무시하고 상체를 일으켰습니다."
   },
   {
    "label": "B",
    "score": 9,
    "verdict_ko": "이전 샷의 누워있는 자세를 정확히 유지한 채, 빈 목덜미를 확인하는 손과 다급한 표정까지 프롬프트의 요구사항을 완벽하게 충족했습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "누워있는 자세를 유지하지 않았으며, 허공에 떠 있는 왼손의 기형적인 묘사로 인해 신체 해부학적 심각한 오류(Hard Violation)가 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S36sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "목을 짚고 있는 손의 손가락(특히 검지)이 비정상적으로 길고 관절이 없어 부자연스럽습니다.",
     "fix_en": "Redraw the hand touching the neck so the fingers have natural anatomical proportions, distinct joints, and realistic knuckles.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the hand touching the neck so the fingers have natural anatomical proportions, distinct joints, and realistic knuckles.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S36sh1"
  }
 },
 "S36sh5": {
  "input_fingerprint": "377d6e72477363b0",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, warm sunlight.\n\nSHOT TEXT (authoritative, Korean): 반쯤 열린 미닫이문을 한 손으로 잡은 채 활짝 웃고 있는 영란의 정지된 찰나.\n\nLOCATION (lock): At the half-open sliding doorway of the restaurant's rear bedroom, with warm daylight entering through the room window. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the tidy back room, half-open sliding door, bedding, and warm daylight. Exclude the woman's neck-checking action and show the elderly woman smiling in the doorway.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 5,
   "A": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "영란이 문을 잡고 웃는 찰나를 정확히 연출했으며, 배경과 인물의 연속성도 잘 유지했습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "영란의 행동은 일치하나, 현대 배경임에도 지시 없는 한복을 입고 있어 연속성을 위반했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "누워있어야 할 지후가 영란 대신 문을 잡고 있어 인물과 행동의 연결이 완전히 어긋났습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S36sh4_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1225321>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "배경의 여성이 여전히 손으로 목을 짚고 있습니다. 지시사항은 여성의 목을 확인하는 행동을 제외할 것을 명시했습니다.",
     "fix_en": "Remove the young woman's hand from her neck, placing her arm resting naturally down by her side on the bed.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the young woman's hand from her neck, placing her arm resting naturally down by her side on the bed.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S36sh4"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S37sh4::bgfirst_bg": {
  "input_fingerprint": "5c6ba5d53b1b1a02",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 흔적도 없는 제단 터를 보며 이해할 수 없다는 듯 미간을 찌푸린 김지후(강수리영)의 측면.\n\nLOCATION (lock): In the restaurant's rear yard at the open doorway of an ordinary storage shed.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level beside 김지후(강수리영), the tracking arc settles into a close profile at the open storage doorway, with her face occupying the right half and the ordinary storage interior remaining legible behind her on the left. Her knitted brow and arrested forward lean hold the frame while the empty area where she expected the altar stays within her sightline, preparing the pivot toward the off-screen voice.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 창고 출입구 (open) — The open front of the doorway faces the camera at an oblique angle; used as The doorway frames the profile and reveals the mundane interior that contradicts her memory; 식당 물품과 낚시 용품 (piled inside the storage space); used as Visible behind her as evidence that the expected altar has vanished; 제단이 있었을 것으로 여긴 자리 (no altar or trace visible); used as A conspicuously ordinary gap held near her line of sight.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 흔적도 없는 제단 터를 보며 이해할 수 없다는 듯 미간을 찌푸린 김지후(강수리영)의 측면.\n\nLOCATION (lock): In the restaurant's rear yard at the open doorway of an ordinary storage shed.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level beside 김지후(강수리영), the tracking arc settles into a close profile at the open storage doorway, with her face occupying the right half and the ordinary storage interior remaining legible behind her on the left. Her knitted brow and arrested forward lean hold the frame while the empty area where she expected the altar stays within her sightline, preparing the pivot toward the off-screen voice.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 창고 출입구 (open) — The open front of the doorway faces the camera at an oblique angle; used as The doorway frames the profile and reveals the mundane interior that contradicts her memory; 식당 물품과 낚시 용품 (piled inside the storage space); used as Visible behind her as evidence that the expected altar has vanished; 제단이 있었을 것으로 여긴 자리 (no altar or trace visible); used as A conspicuously ordinary gap held near her line of sight.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S37sh4__bgfirst_bg.png",
  "asset_id": "e93b6746-970f-4b31-ad6c-ae908c27276a",
  "input_asset_ids": [
   "c0c34a23-8c57-45ff-9764-408401aa2f2c",
   "9379f746-9fec-4d7c-bca5-94d3e636b113"
  ]
 },
 "S37sh4": {
  "input_fingerprint": "5cb20c8b8b325ecc",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 흔적도 없는 제단 터를 보며 이해할 수 없다는 듯 미간을 찌푸린 김지후(강수리영)의 측면.\n\nLOCATION (lock): In the restaurant's rear yard at the open doorway of an ordinary storage shed. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The shrine, goat head, blood bowl, candles, and antler-tipped spear seen here at night have all been removed without a trace. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 흔적도 없는 제단 터를 보며 이해할 수 없다는 듯 미간을 찌푸린 김지후(강수리영)의 측면.\n\nLOCATION (lock): In the restaurant's rear yard at the open doorway of an ordinary storage shed. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The shrine, goat head, blood bowl, candles, and antler-tipped spear seen here at night have all been removed without a trace. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 흔적도 없는 제단 터를 보며 이해할 수 없다는 듯 미간을 찌푸린 김지후(강수리영)의 측면.\n\nLOCATION (lock): In the restaurant's rear yard at the open doorway of an ordinary storage shed. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The shrine, goat head, blood bowl, candles, and antler-tipped spear seen here at night have all been removed without a trace. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S37sh4__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S37sh4.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L30B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "빈 공간을 보며 미간을 찌푸린 측면 모습을 정확한 표정과 적절한 프레이밍으로 잘 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "측면 구도이긴 하나 요구된 찌푸린 표정이 거의 드러나지 않으며 문틀 구조가 레퍼런스와 다릅니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "빈 공간을 보며 미간을 찌푸린 측면 모습을 정확한 표정과 적절한 프레이밍으로 잘 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "측면 구도이긴 하나 요구된 찌푸린 표정이 거의 드러나지 않으며 문틀 구조가 레퍼런스와 다릅니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 측면 구도와 미간을 찌푸린 표정을 훌륭하게 연출하여 지시문의 핵심을 가장 잘 충족함."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "측면 구도는 맞으나 요구된 미간을 찌푸린 표정이 명확히 드러나지 않아 덜 효과적임."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지정된 측면 구도와 미간을 찌푸린 표정을 훌륭하게 연출하여 지시문의 핵심을 가장 잘 충족함."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "측면 구도는 맞으나 요구된 미간을 찌푸린 표정이 명확히 드러나지 않아 덜 효과적임."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 10
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 15,
   "B": 10
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "빈 공간을 보며 미간을 찌푸린 측면 모습을 정확한 표정과 적절한 프레이밍으로 잘 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "측면 구도이긴 하나 요구된 찌푸린 표정이 거의 드러나지 않으며 문틀 구조가 레퍼런스와 다릅니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L30B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지후의 펜던트가 사라진 상태여야 한다는 지시와 달리, 목에 목걸이 줄이 그대로 남아있습니다.",
     "fix_en": "Remove the dark necklace string from around the character's neck.",
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the dark necklace string from around the character's neck.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S37sh4__bgfirst_bg.png",
   "bg_asset_id": "e93b6746-970f-4b31-ad6c-ae908c27276a",
   "bg_record_key": "S37sh4::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S37sh5::bgfirst_bg": {
  "input_fingerprint": "04c389e545bc0b2a",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 식당 뒷문 쪽에 굳은 표정으로 서 있는 강영란(흰 머리 여자)의 전신.\n\nLOCATION (lock): At the restaurant's rear doorway overlooking the daylight yard and open storage shed.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From eye level near the storage doorway, the pan finishes across the yard in a wide full-body reveal of 강영란(흰 머리 여자) at the restaurant rear doorway. She is held slightly off-center within the doorway while the open yard creates an uncomfortable span between her and the unseen 김지후(강수리영), leaving lateral space for the next move onto their dialogue axis.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 식당 뒷문 (occupied by Youngran) — The doorway faces across the yard toward the camera; its open or closed state is unspecified; used as Frames Youngran's full figure and fixes her position across the yard; 식당 뒤뜰 (visible between the two positions); used as Maintains the physical separation between Youngran and the storage-side camera position; 창고 출입구 (open) — Its front is seen obliquely at the edge of the composition; used as Marks the camera's originating side of the reveal.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 식당 뒷문 쪽에 굳은 표정으로 서 있는 강영란(흰 머리 여자)의 전신.\n\nLOCATION (lock): At the restaurant's rear doorway overlooking the daylight yard and open storage shed.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From eye level near the storage doorway, the pan finishes across the yard in a wide full-body reveal of 강영란(흰 머리 여자) at the restaurant rear doorway. She is held slightly off-center within the doorway while the open yard creates an uncomfortable span between her and the unseen 김지후(강수리영), leaving lateral space for the next move onto their dialogue axis.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 식당 뒷문 (occupied by Youngran) — The doorway faces across the yard toward the camera; its open or closed state is unspecified; used as Frames Youngran's full figure and fixes her position across the yard; 식당 뒤뜰 (visible between the two positions); used as Maintains the physical separation between Youngran and the storage-side camera position; 창고 출입구 (open) — Its front is seen obliquely at the edge of the composition; used as Marks the camera's originating side of the reveal.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S37sh5__bgfirst_bg.png",
  "asset_id": "dd848aee-0739-48d1-9e8a-a0aad4ba50a2",
  "input_asset_ids": [
   "e58bfe7d-8c6b-48b6-983f-3dbb1ebc81eb",
   "9379f746-9fec-4d7c-bca5-94d3e636b113"
  ]
 },
 "S37sh5": {
  "input_fingerprint": "e1aa59a2e87ba340",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 식당 뒷문 쪽에 굳은 표정으로 서 있는 강영란(흰 머리 여자)의 전신.\n\nLOCATION (lock): At the restaurant's rear doorway overlooking the daylight yard and open storage shed. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The shed remains an ordinary storage space with restaurant and fishing supplies; the night shrine is absent. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 식당 뒷문 쪽에 굳은 표정으로 서 있는 강영란(흰 머리 여자)의 전신.\n\nLOCATION (lock): At the restaurant's rear doorway overlooking the daylight yard and open storage shed. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The shed remains an ordinary storage space with restaurant and fishing supplies; the night shrine is absent. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 식당 뒷문 쪽에 굳은 표정으로 서 있는 강영란(흰 머리 여자)의 전신.\n\nLOCATION (lock): At the restaurant's rear doorway overlooking the daylight yard and open storage shed. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The shed remains an ordinary storage space with restaurant and fishing supplies; the night shrine is absent. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S37sh5__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S37sh5.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:867030>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L30B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:867030>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시된 전신(Full body) 샷의 앵글과 굳은 표정, 캐릭터의 외양 및 복장을 충실하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "전신 샷 요구사항을 무시하고 무릎 위에서 프레임을 잘랐으며, 레퍼런스의 복장과도 일치하지 않습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시된 전신(Full body) 샷의 앵글과 굳은 표정, 캐릭터의 외양 및 복장을 충실하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "전신 샷 요구사항을 무시하고 무릎 위에서 프레임을 잘랐으며, 레퍼런스의 복장과도 일치하지 않습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "전신(풀샷) 촬영 지시를 어기고 피사체의 하반신을 잘라냈으며, 캐릭터의 의상이 레퍼런스와 일치하지 않습니다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "요청된 전신 프레이밍을 정확히 구현했고, 캐릭터의 외형과 의상, 그리고 지정된 장소와 굳은 표정을 훌륭하게 반영했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "전신(풀샷) 촬영 지시를 어기고 피사체의 하반신을 잘라냈으며, 캐릭터의 의상이 레퍼런스와 일치하지 않습니다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "요청된 전신 프레이밍을 정확히 구현했고, 캐릭터의 외형과 의상, 그리고 지정된 장소와 굳은 표정을 훌륭하게 반영했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 8
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 15,
   "B": 8
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지시된 전신(Full body) 샷의 앵글과 굳은 표정, 캐릭터의 외양 및 복장을 충실하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "전신 샷 요구사항을 무시하고 무릎 위에서 프레임을 잘랐으며, 레퍼런스의 복장과도 일치하지 않습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L30B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:867030>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "인물이 레이아웃 스케치에 지정된 위치(맞은편 흰색 건물의 미닫이문 안쪽)가 아닌 마당 한가운데에 잘못 배치되었습니다.",
     "fix_en": "Move the woman back to stand inside the open sliding doorway of the white building in the background, scaling her down to match the exact position in the layout sketch.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "레퍼런스 이미지에 있는 인물의 머리 장식(전통 머리띠와 장신구)이 누락되었습니다.",
     "fix_en": "Add the traditional headband and ornament to the woman's hair exactly as shown in her character reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Move the woman back to stand inside the open sliding doorway of the white building in the background, scaling her down to match the exact position in the layout sketch.\n- Add the traditional headband and ornament to the woman's hair exactly as shown in her character reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S37sh5__bgfirst_bg.png",
   "bg_asset_id": "dd848aee-0739-48d1-9e8a-a0aad4ba50a2",
   "bg_record_key": "S37sh5::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S37sh10": {
  "input_fingerprint": "e6a23b0577bff4f3",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 꿰뚫어 보는 듯한 강영란(흰 머리 여자)의 날카로운 시선에 압도되어 옴짝달싹 못하고 굳어버린 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): In the rear yard between the restaurant's back door and the open storage shed. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the daylight backyard, open storage shed, restaurant back door, and orderly supplies. Exclude the earlier full-body doorway framing and focus on the younger woman's intimidated expression.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "지시된 클로즈업 앵글과 굳어버린 표정을 잘 구현했으며, 레퍼런스의 배경(선반의 붉은 바구니, 외부의 흰색 건물)을 가장 정확히 재현했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "표정과 앵글은 적절하나, 레퍼런스의 배경 디테일(외부 흰색 건물, 선반 위 특정 소품 등)이 다소 부정확하게 묘사되었습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "바닥에 비정상적인 텍스처 패턴이 생성되었으며, 배경 구조가 지정된 레퍼런스와 전혀 일치하지 않습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S37sh5_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트는 '굳어버린' 표정을 요구하지만, 지후의 얼굴에 눈물이 흐르고 있습니다.",
     "fix_en": "Remove the tears and tear streaks from the character's face.",
     "unfixable": false
    },
    {
     "issue_ko": "이전 샷의 시점(창고 안에서 밖을 보는 방향)을 배경으로 그대로 재사용하여, 카메라를 보는 지후가 바깥에 있는 강영란을 등지고 창고 안쪽을 바라보는 모순된 공간 배치가 되었습니다.",
     "fix_en": "This spatial geometry error cannot be fixed in-place; the background places the character facing into the shed with her back to the yard.",
     "unfixable": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the tears and tear streaks from the character's face.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S37sh5"
  }
 },
 "S38sh1::bgfirst_bg": {
  "input_fingerprint": "0570d6e90f3c7e45",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 선착장에 정박된 낚시 어선 뱃머리에서 기괴한 종이 인형들이 매달린 새끼줄을 양손으로 쥔 채 고정하고 있는 황봉업(황 씨)의 모습.\n\nLOCATION (lock): At the bow of a fishing boat moored beside the island pier in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height on the dock beside the bow, a static medium view faces inward toward 황봉업(황 씨) as he fixes the rope decoration with both hands. His working arms form a shallow diagonal across the center, while the bow and the hanging paper figures remain proportionate environmental context rather than oversized foreground objects.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 정박한 낚시 어선 뱃머리 (moored at the dock) — The bow points obliquely past the camera from the dock side; used as Supports Huang and establishes his position at the working end of the vessel; 새끼줄 장식 (being secured); used as Primary working detail held between Huang's hands and across the bow; 여러 형상의 종이 인형 (cut from paper and hanging from the rope); used as Creates the uncanny graphic rhythm beneath the rope while retaining visible surrounding context; 선착장 (beside the moored boat) — Its length recedes laterally past the boat; used as Establishes the camera-side platform and the route of the subsequent lateral track.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 선착장에 정박된 낚시 어선 뱃머리에서 기괴한 종이 인형들이 매달린 새끼줄을 양손으로 쥔 채 고정하고 있는 황봉업(황 씨)의 모습.\n\nLOCATION (lock): At the bow of a fishing boat moored beside the island pier in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height on the dock beside the bow, a static medium view faces inward toward 황봉업(황 씨) as he fixes the rope decoration with both hands. His working arms form a shallow diagonal across the center, while the bow and the hanging paper figures remain proportionate environmental context rather than oversized foreground objects.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 정박한 낚시 어선 뱃머리 (moored at the dock) — The bow points obliquely past the camera from the dock side; used as Supports Huang and establishes his position at the working end of the vessel; 새끼줄 장식 (being secured); used as Primary working detail held between Huang's hands and across the bow; 여러 형상의 종이 인형 (cut from paper and hanging from the rope); used as Creates the uncanny graphic rhythm beneath the rope while retaining visible surrounding context; 선착장 (beside the moored boat) — Its length recedes laterally past the boat; used as Establishes the camera-side platform and the route of the subsequent lateral track.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S38sh1__bgfirst_bg.png",
  "asset_id": "18656478-9287-43f9-b18a-9501a274a334",
  "input_asset_ids": [
   "b2734578-fd5a-4a9b-96d2-fe540ee81290",
   "6aadab04-f42a-47e3-8015-09aca9fa1ea5",
   "42e89fec-1bac-413f-a9c4-449f887fe56d"
  ]
 },
 "S38sh1": {
  "input_fingerprint": "f9b7ed9216873930",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 선착장에 정박된 낚시 어선 뱃머리에서 기괴한 종이 인형들이 매달린 새끼줄을 양손으로 쥔 채 고정하고 있는 황봉업(황 씨)의 모습.\n\nLOCATION (lock): At the bow of a fishing boat moored beside the island pier in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The fishing boat remains decorated with a straw rope bearing cut-paper figures.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 선착장에 정박된 낚시 어선 뱃머리에서 기괴한 종이 인형들이 매달린 새끼줄을 양손으로 쥔 채 고정하고 있는 황봉업(황 씨)의 모습.\n\nLOCATION (lock): At the bow of a fishing boat moored beside the island pier in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The fishing boat remains decorated with a straw rope bearing cut-paper figures.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 선착장에 정박된 낚시 어선 뱃머리에서 기괴한 종이 인형들이 매달린 새끼줄을 양손으로 쥔 채 고정하고 있는 황봉업(황 씨)의 모습.\n\nLOCATION (lock): At the bow of a fishing boat moored beside the island pier in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The fishing boat remains decorated with a straw rope bearing cut-paper figures.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S38sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S38sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L23B01.png"
    },
    {
     "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_restaurant_property_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "인물이 뱃머리에서 새끼줄을 쥐고 있는 구도와 기괴한 종이 인형의 디테일은 매우 우수하나, 필수 지정된 구조물 외관(Structure Look) 레퍼런스 반영에 실패했습니다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "지정된 구조물 외관은 잘 반영했으나, 인물이 뱃머리가 아닌 선착장에 서 있어 텍스트의 핵심 행동 위치를 위반했으며 종이 인형의 기괴한 특징도 누락되었습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "인물이 뱃머리에서 새끼줄을 쥐고 있는 구도와 기괴한 종이 인형의 디테일은 매우 우수하나, 필수 지정된 구조물 외관(Structure Look) 레퍼런스 반영에 실패했습니다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "지정된 구조물 외관은 잘 반영했으나, 인물이 뱃머리가 아닌 선착장에 서 있어 텍스트의 핵심 행동 위치를 위반했으며 종이 인형의 기괴한 특징도 누락되었습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "지정된 건축물 레퍼런스(석재 벽, 목재 데크 등)를 배경에 정확히 반영했으나, 인물이 어선 뱃머리가 아닌 선착장에 서 있는 점과 종이 인형의 기괴한 묘사가 부족한 점이 감점 요인입니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "어선 뱃머리에 선 인물의 위치와 기괴한 종이 인형의 묘사는 프롬프트에 부합하지만, 절대적인 우선순위인 '건축물 레퍼런스'를 완전히 무시하고 원본 배경 사진의 건물을 그대로 렌더링하여 우선순위 3을 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "지정된 건축물 레퍼런스(석재 벽, 목재 데크 등)를 배경에 정확히 반영했으나, 인물이 어선 뱃머리가 아닌 선착장에 서 있는 점과 종이 인형의 기괴한 묘사가 부족한 점이 감점 요인입니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "어선 뱃머리에 선 인물의 위치와 기괴한 종이 인형의 묘사는 프롬프트에 부합하지만, 절대적인 우선순위인 '건축물 레퍼런스'를 완전히 무시하고 원본 배경 사진의 건물을 그대로 렌더링하여 우선순위 3을 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 11
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 11,
   "B": 11
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "인물이 뱃머리에서 새끼줄을 쥐고 있는 구도와 기괴한 종이 인형의 디테일은 매우 우수하나, 필수 지정된 구조물 외관(Structure Look) 레퍼런스 반영에 실패했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "지정된 구조물 외관은 잘 반영했으나, 인물이 뱃머리가 아닌 선착장에 서 있어 텍스트의 핵심 행동 위치를 위반했으며 종이 인형의 기괴한 특징도 누락되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L23B01.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_restaurant_property_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "남자가 종이 인형이 매달린 새끼줄이 아닌 얇은 일반 줄과 막대기 같은 형태를 쥐고 있습니다.",
     "fix_en": "Replace the thin string and stick-like object in the man's hands with the thick straw rope that has paper dolls hanging from it.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "줄을 쥐고 있는 화면 앞쪽(왼손)의 손가락들이 뭉개져 해부학적으로 어색합니다.",
     "fix_en": "Redraw the foreground hand with distinct, anatomically correct fingers naturally gripping the rope.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the thin string and stick-like object in the man's hands with the thick straw rope that has paper dolls hanging from it.\n- Redraw the foreground hand with distinct, anatomically correct fingers naturally gripping the rope.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S38sh1__bgfirst_bg.png",
   "bg_asset_id": "18656478-9287-43f9-b18a-9501a274a334",
   "bg_record_key": "S38sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate",
   "seed_attached": true
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S38sh7": {
  "input_fingerprint": "7482d399acc21d58",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 마을 청년 #1과 마을 청년 #2가 김지후(강수리영)를 향해 사람 좋은 환한 미소를 짓는 모습.\n\nLOCATION (lock): On the island pier beside the decorated moored fishing boat. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the sunlit pier, moored fishing boat, rope decorations, and harbor backdrop. Exclude the boatman's solitary decorating pose and include only the two smiling youths near the woman.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing; the two youths are the same men who pursued her from the slaughterhouse.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 5,
   "C": 2
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "두 남성이 여성을 향해 밝게 웃는 모습을 자연스럽게 연출했으며, 이전 장면의 뱃사공 캐릭터를 적절히 제외하고 새로운 인물들로 구성했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "요구된 샷의 구도와 미소는 반영되었으나, 배제해야 할 이전 장면의 뱃사공과 동일한 복장의 인물이 그대로 등장해 지시를 완전히 따르지 못했습니다."
   },
   {
    "label": "C",
    "score": 2,
    "verdict_ko": "왼쪽 남성의 오른쪽 어깨 위에 해부학적으로 설명할 수 없는 정체불명의 손이 나타나 하드 위반입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S38sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "두 남성의 외모가 40~50대 중년으로 묘사되어 있어, 샷 텍스트에 명시된 '청년'이라는 설정에 부합하지 않습니다.",
     "fix_en": "Rejuvenate the faces and skin of the two men so they appear as young men in their 20s.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Rejuvenate the faces and skin of the two men so they appear as young men in their 20s.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S38sh1"
  }
 },
 "groupbg::섬마을_골목": {
  "input_fingerprint": "81a9c98986e07e78",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "섬마을_골목",
    "tags": [
     "S22sh2",
     "S39sh3",
     "S39sh5",
     "S39sh7"
    ]
   },
   "context_sig": "70fa36c6dab4d7f6"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On a daylight lane through the island's coastal village, beside low walls and house gardens.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인) / 반월 식당 내부: 초등학교 교실 크기에 식당, 편의점, 낚시 용품점이 섞인 잡화점 느낌의 식당. 나무 테이블과 주방, 진열대가 혼재되어 있다. (특징: 다수의 낡은 목재 테이블; 격자무늬 나무 유리창; 개방된 주방 공간; 낚시 용품과 잡화 진열대; 붉은 피가 흐르는 뭉티기와 생간 등 날고기 한 상)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 일각 - 낮/ 밤\n- 마을 곳곳에 생생하게 폈던 꽃들도 사르륵 사그라져 시들어버린다.\n- 실외. 금월도 해안 마을 일각 - 낮\n- 집 앞 대문에 황 씨가 배에 걸었던 것과 비슷한 새끼줄 장식을 걸던 주민들\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On a daylight lane through the island's coastal village, beside low walls and house gardens.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인) / 반월 식당 내부: 초등학교 교실 크기에 식당, 편의점, 낚시 용품점이 섞인 잡화점 느낌의 식당. 나무 테이블과 주방, 진열대가 혼재되어 있다. (특징: 다수의 낡은 목재 테이블; 격자무늬 나무 유리창; 개방된 주방 공간; 낚시 용품과 잡화 진열대; 붉은 피가 흐르는 뭉티기와 생간 등 날고기 한 상)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 일각 - 낮/ 밤\n- 마을 곳곳에 생생하게 폈던 꽃들도 사르륵 사그라져 시들어버린다.\n- 실외. 금월도 해안 마을 일각 - 낮\n- 집 앞 대문에 황 씨가 배에 걸었던 것과 비슷한 새끼줄 장식을 걸던 주민들\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_섬마을_골목_f099b6.png",
  "asset_id": "c5817d07-42e5-46ac-9251-888fccb5ce2d",
  "input_asset_ids": [
   "f5438b8d-9573-469e-90ca-444e5b3ea4de"
  ],
  "origin_tag": "S39sh3",
  "place_text": "On a daylight lane through the island's coastal village, beside low walls and house gardens.",
  "origin_inputs": {
   "place_text": "On a daylight lane through the island's coastal village, beside low walls and house gardens.",
   "time_of_day_en": "day",
   "conti_asset_id": "f5438b8d-9573-469e-90ca-444e5b3ea4de"
  }
 },
 "S39sh3::bgfirst_bg": {
  "input_fingerprint": "9b931e7202e2787c",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 훅 끼쳐온 역겨운 냄새에 코를 움켜쥔 채 미간을 잔뜩 구기고 있는 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): On a daylight lane through the island's coastal village, beside low walls and house gardens.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level directly ahead and slightly off-axis, the dolly-in ends on a close three-quarter view emphasizing 김지후(강수리영)'s pinched nose, tightly furrowed brow, and compressed mouth without excluding the hand causing the gesture. Her face and hand take most of the frame, with only a narrow strip of the village road retained as spatial context.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 마을 길 (partially visible); used as Soft contextual strip behind the close reaction, preserving her location and direction of travel; 마을 집들 (soft in the background) — Only partial house fronts are visible behind her; used as An indistinct background cue that anticipates the following reveal without competing with her expression.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 훅 끼쳐온 역겨운 냄새에 코를 움켜쥔 채 미간을 잔뜩 구기고 있는 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): On a daylight lane through the island's coastal village, beside low walls and house gardens.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level directly ahead and slightly off-axis, the dolly-in ends on a close three-quarter view emphasizing 김지후(강수리영)'s pinched nose, tightly furrowed brow, and compressed mouth without excluding the hand causing the gesture. Her face and hand take most of the frame, with only a narrow strip of the village road retained as spatial context.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 마을 길 (partially visible); used as Soft contextual strip behind the close reaction, preserving her location and direction of travel; 마을 집들 (soft in the background) — Only partial house fronts are visible behind her; used as An indistinct background cue that anticipates the following reveal without competing with her expression.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S39sh3__bgfirst_bg.png",
  "asset_id": "9309f0c1-a35c-4ed7-b9f7-ce8dc7c7fb5a",
  "input_asset_ids": [
   "f5438b8d-9573-469e-90ca-444e5b3ea4de",
   "c5817d07-42e5-46ac-9251-888fccb5ce2d"
  ]
 },
 "S39sh3": {
  "input_fingerprint": "01e3c8ac8db91d62",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 훅 끼쳐온 역겨운 냄새에 코를 움켜쥔 채 미간을 잔뜩 구기고 있는 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): On a daylight lane through the island's coastal village, beside low walls and house gardens. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 훅 끼쳐온 역겨운 냄새에 코를 움켜쥔 채 미간을 잔뜩 구기고 있는 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): On a daylight lane through the island's coastal village, beside low walls and house gardens. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 훅 끼쳐온 역겨운 냄새에 코를 움켜쥔 채 미간을 잔뜩 구기고 있는 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): On a daylight lane through the island's coastal village, beside low walls and house gardens. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S39sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S39sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_섬마을_골목_f099b6.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지정된 얼굴 클로즈업 샷과 표정을 잘 구현했으며, 레퍼런스의 의상(초록색 스웨터)과 배경을 정확히 유지했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "클로즈업 샷의 구도와 표정은 좋으나, 캐릭터 레퍼런스의 의상(스웨터)을 전혀 다른 옷으로 임의 변경하여 감점되었습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지정된 얼굴 클로즈업 샷과 표정을 잘 구현했으며, 레퍼런스의 의상(초록색 스웨터)과 배경을 정확히 유지했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "클로즈업 샷의 구도와 표정은 좋으나, 캐릭터 레퍼런스의 의상(스웨터)을 전혀 다른 옷으로 임의 변경하여 감점되었습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지문이 요구한 '얼굴 클로즈업' 샷 사이즈에 완벽히 부합하며, 코를 쥐고 미간을 찌푸린 표정 연기를 탁월하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "행동과 표정 연기는 적절히 묘사되었으나, 프레임이 다소 넓어져 '얼굴 클로즈업'이라는 핵심 샷 지시에서 멀어졌습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지문이 요구한 '얼굴 클로즈업' 샷 사이즈에 완벽히 부합하며, 코를 쥐고 미간을 찌푸린 표정 연기를 탁월하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "행동과 표정 연기는 적절히 묘사되었으나, 프레임이 다소 넓어져 '얼굴 클로즈업'이라는 핵심 샷 지시에서 멀어졌습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 12
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 12
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "지정된 얼굴 클로즈업 샷과 표정을 잘 구현했으며, 레퍼런스의 의상(초록색 스웨터)과 배경을 정확히 유지했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "클로즈업 샷의 구도와 표정은 좋으나, 캐릭터 레퍼런스의 의상(스웨터)을 전혀 다른 옷으로 임의 변경하여 감점되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_섬마을_골목_f099b6.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "시선이 스케치의 화살표 방향(오른쪽 배경)을 따르지 않고 아래쪽 도로를 향하고 있습니다.",
     "fix_en": "Adjust her eyes to look towards the right background, following the exact direction of the arrow in the layout sketch.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "코를 쥐고 있는 손의 해부학적 구조가 어색하며(오른팔 각도에 맞지 않는 손가락 배치), 스케치의 손 모양과 다릅니다.",
     "fix_en": "Redraw the right hand to anatomically correctly pinch the nose, positioning the thumb on the side nearest the camera and curling the fingers over the front as shown in the sketch.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Adjust her eyes to look towards the right background, following the exact direction of the arrow in the layout sketch.\n- Redraw the right hand to anatomically correctly pinch the nose, positioning the thumb on the side nearest the camera and curling the fingers over the front as shown in the sketch.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S39sh3__bgfirst_bg.png",
   "bg_asset_id": "9309f0c1-a35c-4ed7-b9f7-ce8dc7c7fb5a",
   "bg_record_key": "S39sh3::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "섬마을_골목",
   "groupbg_asset_id": "c5817d07-42e5-46ac-9251-888fccb5ce2d"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S39sh5": {
  "input_fingerprint": "ec670fccc2f53f47",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 집집마다 널려 있는 무성한 녹색 석창포 다발들을 멍하니 응시하며 멈춰 서 있는 김지후(강수리영)의 모습.\n\nLOCATION (lock): Along the coastal village lane where bundles of green medicinal plants fill the areas outside the houses. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the sunny village lane, low houses, stone walls, and abundant green herb bundles. Exclude the earlier face-only smell reaction and show the woman stopped before the plants.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Bundles or plantings of green seokchangpo remain present at the village houses. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "이전 샷의 장소(검은 기와집, 돌담, 전봇대)를 정확히 유지하면서, 식물을 멍하니 응시하는 인물의 모습을 훌륭하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "레퍼런스에 없던 파란 지붕 등 배경 건축물이 완전히 변경되어 장소 일치(Location lock) 조건에 어긋납니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "장소의 왼쪽 형태는 비슷하나, 도로 오른쪽에 원본에 없는 높은 돌담이 생겨 공간의 연속성이 훼손되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S39sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "스웨터 목선 안쪽에 흰색 셔츠가 노출되어 있습니다. 레퍼런스에 있는 어두운 올리브색 셔츠와 일치하지 않습니다.",
     "fix_en": "Change the white undershirt visible at the neckline to dark olive green.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the white undershirt visible at the neckline to dark olive green.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S39sh3"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S39sh7": {
  "input_fingerprint": "3576d9c19176467d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 낮은 돌담 너머로 일제히 고개를 내민 채 김지후(강수리영)를 구경하듯 응시하고 있는 금월도 마을 주민들의 모습.\n\nLOCATION (lock): Beside the low stone walls lining the island village lane. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the village lane, low stone wall, house fronts, herb plants, and daylight. Exclude the woman's isolated reaction and show residents peering over the wall.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 청년 (한국인 20~30대 남성, 짧은 검은 머리, 그을린 얼굴, 거친 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 3,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "돌담 너머로 쳐다보는 주민들의 모습을 참조된 배경 구조에 정확히 맞춰 연출했으나, 지정된 청년 캐릭터 외의 다양한 인물들이 포함되어 엄격한 캐릭터 제한 지침에서 감점되었습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "원래 도로가 있어야 할 위치를 가로막는 돌담이 생성되어 공간 구조가 파괴되었으며, 허용되지 않은 인물들이 다수 등장해 지침을 심각하게 위반했습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "식물 건조대와 담장의 위치가 참조 이미지의 반대편으로 뒤집혀 장소의 일관성이 어긋났으며, 제시된 청년 이외의 인물들이 추가되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S39sh5_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 청년: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:756328>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "인물 제약 조건에 어긋나는 노인 여성과 중장년 주민 3명이 프레임에 포함되어 있습니다 ('금월도 마을 청년'만 등장 가능).",
     "fix_en": "Replace the older woman and the three other older residents peering over the wall with additional instances of the young man in the green beanie and navy jacket.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "청년이 대문 위에 얹은 손의 형태가 심하게 뭉개져 기괴한 덩어리처럼 보입니다.",
     "fix_en": "Redraw the young man's hand resting on the wooden gate into a properly articulated hand wearing the bright orange and blue rubber work glove from the character reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the older woman and the three other older residents peering over the wall with additional instances of the young man in the green beanie and navy jacket.\n- Redraw the young man's hand resting on the wooden gate into a properly articulated hand wearing the bright orange and blue rubber work glove from the character reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S39sh5"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S40sh1::bgfirst_bg": {
  "input_fingerprint": "9f35137571594ee4",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 대낮임에도 불구하고 어둠이 짙게 깔린 숲길 오솔길 입구 앞에 우두커니 선 김지후(강수리영)의 뒷모습.\n\nLOCATION (lock): At the entrance of a narrow forest trail above the coastal village, where daylight fades into dense shade.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From standing height several meters behind 김지후(강수리영), the dolly-in begins on a centered wide view aligned with her back and the forest path entrance. Her full figure occupies the lower center while the path continues beyond her into the darker interior, turning her stillness into a threshold she has not yet crossed.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the lower-center of the frame, midground; 숲 오솔길 입구와 이어지는 길 in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 숲으로 통하는 오솔길 (extending into the forest); used as Central depth axis that draws the eye past Suriyoung and into the forest; 숲 입구 (darker beyond despite daytime) — The opening faces the camera along the same axis as Suriyoung's back; used as Frames the exact threshold at which Suriyoung has stopped; 숲 (light does not appear to reach the deeper area); used as Encloses the route and limits lateral escape in the composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 대낮임에도 불구하고 어둠이 짙게 깔린 숲길 오솔길 입구 앞에 우두커니 선 김지후(강수리영)의 뒷모습.\n\nLOCATION (lock): At the entrance of a narrow forest trail above the coastal village, where daylight fades into dense shade.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From standing height several meters behind 김지후(강수리영), the dolly-in begins on a centered wide view aligned with her back and the forest path entrance. Her full figure occupies the lower center while the path continues beyond her into the darker interior, turning her stillness into a threshold she has not yet crossed.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the lower-center of the frame, midground; 숲 오솔길 입구와 이어지는 길 in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 숲으로 통하는 오솔길 (extending into the forest); used as Central depth axis that draws the eye past Suriyoung and into the forest; 숲 입구 (darker beyond despite daytime) — The opening faces the camera along the same axis as Suriyoung's back; used as Frames the exact threshold at which Suriyoung has stopped; 숲 (light does not appear to reach the deeper area); used as Encloses the route and limits lateral escape in the composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S40sh1__bgfirst_bg.png",
  "asset_id": "b70cead8-5fda-450d-bebe-69ce64f8b546",
  "input_asset_ids": [
   "2163dc8a-f9c0-4223-b35c-80884fde7f97",
   "09cf0b8b-8857-49e2-af2b-fcea7222c63c"
  ]
 },
 "S40sh1": {
  "input_fingerprint": "eb8f7b7b2b4b9ccb",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 대낮임에도 불구하고 어둠이 짙게 깔린 숲길 오솔길 입구 앞에 우두커니 선 김지후(강수리영)의 뒷모습.\n\nLOCATION (lock): At the entrance of a narrow forest trail above the coastal village, where daylight fades into dense shade. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 대낮임에도 불구하고 어둠이 짙게 깔린 숲길 오솔길 입구 앞에 우두커니 선 김지후(강수리영)의 뒷모습.\n\nLOCATION (lock): At the entrance of a narrow forest trail above the coastal village, where daylight fades into dense shade. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 대낮임에도 불구하고 어둠이 짙게 깔린 숲길 오솔길 입구 앞에 우두커니 선 김지후(강수리영)의 뒷모습.\n\nLOCATION (lock): At the entrance of a narrow forest trail above the coastal village, where daylight fades into dense shade. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S40sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S40sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송숲_오솔길_22bdfa.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "인물의 뒷모습을 화면에 적절한 크기로 배치하여 텍스트의 요구사항(우두커니 선 뒷모습)을 영화적 구도로 잘 살렸습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "지정된 장소와 인물의 뒷모습은 반영되었으나, 로케이션 레퍼런스의 구도를 그대로 답습하여 피사체의 비중이 다소 약합니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "인물의 뒷모습을 화면에 적절한 크기로 배치하여 텍스트의 요구사항(우두커니 선 뒷모습)을 영화적 구도로 잘 살렸습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "지정된 장소와 인물의 뒷모습은 반영되었으나, 로케이션 레퍼런스의 구도를 그대로 답습하여 피사체의 비중이 다소 약합니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "인물의 뒷모습을 적절한 프레이밍으로 포착했으며, 장소 레퍼런스의 자연스러운 음영과 질감을 충실하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "뒷모습 지문은 반영되었으나, 숲길 끝을 부자연스러운 검은 공간으로 처리하여 장소 레퍼런스의 사실성을 훼손했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "인물의 뒷모습을 적절한 프레이밍으로 포착했으며, 장소 레퍼런스의 자연스러운 음영과 질감을 충실하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "뒷모습 지문은 반영되었으나, 숲길 끝을 부자연스러운 검은 공간으로 처리하여 장소 레퍼런스의 사실성을 훼손했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 12,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "인물의 뒷모습을 화면에 적절한 크기로 배치하여 텍스트의 요구사항(우두커니 선 뒷모습)을 영화적 구도로 잘 살렸습니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "지정된 장소와 인물의 뒷모습은 반영되었으나, 로케이션 레퍼런스의 구도를 그대로 답습하여 피사체의 비중이 다소 약합니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송숲_오솔길_22bdfa.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S40sh1__bgfirst_bg.png",
   "bg_asset_id": "b70cead8-5fda-450d-bebe-69ce64f8b546",
   "bg_record_key": "S40sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "해송숲_오솔길",
   "groupbg_asset_id": "09cf0b8b-8857-49e2-af2b-fcea7222c63c"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S40sh8": {
  "input_fingerprint": "dfbc8611a9732b26",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 숲속의 수리영을 공격한 대형 검은 염소와 홀린 듯 시선을 맞춘 채 오솔길 입구에 굳어 있는 김지후(강수리영)의 모습.\n\nLOCATION (lock): At the shaded boundary between the village trail and the dense forest interior. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the shaded pine-trail entrance, bright village-side daylight, and the woman's fixed position. Exclude flashback imagery and add only the large black goat watching from within the woods.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 8,
   "A": 4,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "이전 컷의 고정된 인물 자세를 유지한 채 카메라만 측면으로 이동하여, 숲속의 흑염소와 시선을 교환하는 상황과 마을 쪽의 밝은 배경을 완벽하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "인물이 염소가 아닌 카메라 쪽으로 고개를 돌려 시선 교환 지시를 위반했으며, 이전 컷의 고정된 자세를 유지하지 못했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "인물이 숲이 아닌 정면을 향해 서 있어 자세 고정 규칙을 완전히 위반했으며, 염소와의 시선 교환 및 마을 배경 지시도 누락되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S40sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 컷에서 숲을 향해 똑바로 서 있던 지후의 몸 방향이 마을(오른쪽)을 향한 채 고개만 숲으로 돌린 자세로 변경되어, 인물의 고정된 자세를 유지해야 한다는 규칙을 위반했습니다.",
     "fix_en": "Rotate Jihu's torso, hips, and feet to face the forest (left) so she stands squarely facing the goat, matching her fixed orientation from the previous still.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Rotate Jihu's torso, hips, and feet to face the forest (left) so she stands squarely facing the goat, matching her fixed orientation from the previous still.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S40sh1"
  }
 },
 "groupbg::이름비석_숲길": {
  "input_fingerprint": "1ea5593544bef325",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "이름비석_숲길",
    "tags": [
     "S41sh4",
     "S41sh7",
     "S43sh5",
     "S43sh6",
     "S46sh11",
     "S46sh16",
     "S46sh3",
     "S57sh2",
     "S57sh3",
     "S57sh5"
    ]
   },
   "context_sig": "e52d74aa4aa2d301"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: Beside a moss-covered marker stone deep along the daytime pine-forest path, with rotten meat at its base.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n국도변 주유소: 시골 지방 국도변에 위치한 주유소의 외부 콘크리트 공간. (특징: 낡은 주유기 기둥; 콘크리트 바닥; 한적한 주변 도로 풍경) / 금월도 숲속 입구: 마을에서 산으로 이어지는 숲의 진입로. 숲 안쪽은 그늘져 어둡다. (특징: 좁은 흙길; 입구부터 짙게 드리운 나무 그림자; 마을과 숲의 경계가 되는 수풀) / 금월도 해송 숲길 및 숲속: 해송이 울창하게 우거져 빛이 잘 들지 않는 빽빽하고 좁은 흙길과 숲의 심층부. (특징: 빽빽한 해송 나무 기둥들; 무성하고 어두운 수풀; 바닥에 깔린 흙과 솔잎; ‘수리영’이 적힌 이끼 낀 비석들; 비석 앞에 놓인 썩은 핏빛 육고기 덩어리) / 금월도 해안길: 섬 가장자리를 따라 굽이치는 좁은 길과 가파른 갯바위 지대. (특징: 좁고 험한 해안 흙길; 바다와 인접한 가파르고 미끄러운 바위 지형)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 밤에는 보지 못 했던 허리 높이의 비석들이 눈에 띈다.\n- 비석 앞 썩은 고기의 형태도, 놓여진 상태도, 그대로다.\n- 곳곳에 보이는 몇 개의 비석들에 죄다 “수리영”이란 이름이 새겨져 있다.\n- “수리영” 글자가 흐릿하게 적혀있는 비석.\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: Beside a moss-covered marker stone deep along the daytime pine-forest path, with rotten meat at its base.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n국도변 주유소: 시골 지방 국도변에 위치한 주유소의 외부 콘크리트 공간. (특징: 낡은 주유기 기둥; 콘크리트 바닥; 한적한 주변 도로 풍경) / 금월도 숲속 입구: 마을에서 산으로 이어지는 숲의 진입로. 숲 안쪽은 그늘져 어둡다. (특징: 좁은 흙길; 입구부터 짙게 드리운 나무 그림자; 마을과 숲의 경계가 되는 수풀) / 금월도 해송 숲길 및 숲속: 해송이 울창하게 우거져 빛이 잘 들지 않는 빽빽하고 좁은 흙길과 숲의 심층부. (특징: 빽빽한 해송 나무 기둥들; 무성하고 어두운 수풀; 바닥에 깔린 흙과 솔잎; ‘수리영’이 적힌 이끼 낀 비석들; 비석 앞에 놓인 썩은 핏빛 육고기 덩어리) / 금월도 해안길: 섬 가장자리를 따라 굽이치는 좁은 길과 가파른 갯바위 지대. (특징: 좁고 험한 해안 흙길; 바다와 인접한 가파르고 미끄러운 바위 지형)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 밤에는 보지 못 했던 허리 높이의 비석들이 눈에 띈다.\n- 비석 앞 썩은 고기의 형태도, 놓여진 상태도, 그대로다.\n- 곳곳에 보이는 몇 개의 비석들에 죄다 “수리영”이란 이름이 새겨져 있다.\n- “수리영” 글자가 흐릿하게 적혀있는 비석.\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_이름비석_숲길_b7baf3.png",
  "asset_id": "47256cc0-0523-4a95-8f0e-7b08cbca588a",
  "input_asset_ids": [
   "27c23943-9082-4dbf-9538-a2a5dd89321e"
  ],
  "origin_tag": "S41sh4",
  "place_text": "Beside a moss-covered marker stone deep along the daytime pine-forest path, with rotten meat at its base.",
  "origin_inputs": {
   "place_text": "Beside a moss-covered marker stone deep along the daytime pine-forest path, with rotten meat at its base.",
   "time_of_day_en": "day",
   "conti_asset_id": "27c23943-9082-4dbf-9538-a2a5dd89321e"
  }
 },
 "S41sh4::bgfirst_bg": {
  "input_fingerprint": "9d77a9cf0696a363",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 역겨운 냄새에 미간을 팍 찌푸리며 썩은 고기를 내려다보는 김지후(강수리영)의 얼굴.\n\nLOCATION (lock): Beside a moss-covered marker stone deep along the daytime pine-forest path, with rotten meat at its base.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From shoulder height just in front of and slightly beside 김지후(강수리영), the dolly-in settles into a tight three-quarter face close-up while preserving her lowered eyeline. Her face occupies the upper-center and right of frame, with a restrained sliver of the stone marker and rotten meat remaining below as the cause of her disgust.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 이끼 낀 비석 (오래되고 이끼가 잔뜩 낀 상태) — Its inscribed/front side angles away beneath her lowered face; used as A partial lower-frame reference that connects her expression to what she is examining; 썩은 육고기 (썩은 상태로 비석 앞에 놓여 있음); used as Kept small and slightly soft at the lower edge as the source of her reaction.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "This is a pre-visualization concept art for a film production. Storyboard-quality still frame for the director's shot planning. Cinematic composition, dramatic lighting, professional film production context. NO gore, NO explicit violence, NO blood — focus on dramatic tension and emotion.\n\nCreate ONE PHOTOREALISTIC LIVE-ACTION EMPTY BACKGROUND PLATE for a film shot, 16:9. This must look like a real location photograph, not an illustration, painting, drawing, or 3D render. NO PEOPLE, no figures, no silhouettes, no body parts, no reflections of people, no sketch lines, and no arrows anywhere. No text overlays or watermarks.\n\nREFERENCE USE:\n- Use the FIRST attached thin-line storyboard image ONLY for camera angle, horizon, perspective, and the placement and apparent size of buildings, terrain, and set masses. Ignore all sketched people and arrows entirely.\n- Use the SECOND attached LOCATION PHOTOGRAPH for the real place’s architecture, natural materials, signage, vegetation, and fixed features, re-projecting them into the storyboard camera. If the cameras differ, the storyboard camera wins.\n- Preserve true physical scale using human-scale calibration: a door is approximately 2 m high, a window approximately 1–1.5 m wide, and one storey approximately 2.5–3 m. Never inflate small structures or shrink large ones.\n\nLOCATION LOCK: Daytime on a secluded pine-forest path beside an old stone marker heavily covered with natural moss. At the base of the marker is a small, weathered, opaque food parcel suggesting long-spoiled discarded provisions. The parcel remains closed and non-graphic: no visible meat, flesh, tissue, bones, carcass, blood, bodily remains, stains, wounds, insects, or fluids.\n\nTIME OF DAY LOCK: Day.\n\nCAMERA AND COMPOSITION:\n- Camera at adult shoulder height, positioned just in front of and slightly beside the empty mark where the performer will later stand.\n- Compose this as the clean background plate for a tight three-quarter facial close-up that will be added later. Leave natural negative space in the upper-center and right of frame for the future performer’s face and shoulders, but include no person now.\n- Preserve the intended lowered eyeline through the placement of the practical set elements: a restrained sliver of the moss-covered stone marker enters the lower portion of frame, and its inscribed front plane angles away from the empty performer position.\n- Keep the closed spoiled-food parcel small, unobtrusive, and slightly soft at the lower edge. It should function only as a subtle environmental cue and must not resemble a body part or graphic biological material.\n- The stone and parcel must remain partial lower-frame references rather than dominant foreground objects.\n- Keep realistic physical scale and lens perspective. Background fixtures, trees, and distant objects occupy only the screen area dictated by their actual size and distance.\n\nLIGHTING AND IMAGE QUALITY:\nNatural daytime forest light filtered through pine branches, realistic moss and weathered stone textures, restrained cinematic contrast, shallow depth of field appropriate to a live-action close-up background plate, professional full-frame cinematography, physically accurate optics, subtle filmic color grading, high-detail photorealism. Quiet, tense atmosphere without depicting violence, injury, death, or graphic decay.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S41sh4__bgfirst_bg.png",
  "asset_id": "701789dd-113d-4674-82ff-d5499a345991",
  "input_asset_ids": [
   "27c23943-9082-4dbf-9538-a2a5dd89321e",
   "47256cc0-0523-4a95-8f0e-7b08cbca588a"
  ]
 },
 "S41sh4": {
  "input_fingerprint": "b075dcc657c84ddd",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 역겨운 냄새에 미간을 팍 찌푸리며 썩은 고기를 내려다보는 김지후(강수리영)의 얼굴.\n\nLOCATION (lock): Beside a moss-covered marker stone deep along the daytime pine-forest path, with rotten meat at its base. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The moss-covered marker stones remain standing with rotten meat placed before each one. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 역겨운 냄새에 미간을 팍 찌푸리며 썩은 고기를 내려다보는 김지후(강수리영)의 얼굴.\n\nLOCATION (lock): Beside a moss-covered marker stone deep along the daytime pine-forest path, with rotten meat at its base. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The moss-covered marker stones remain standing with rotten meat placed before each one. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 역겨운 냄새에 미간을 팍 찌푸리며 썩은 고기를 내려다보는 김지후(강수리영)의 얼굴.\n\nLOCATION (lock): Beside a moss-covered marker stone deep along the daytime pine-forest path, with rotten meat at its base. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The moss-covered marker stones remain standing with rotten meat placed before each one. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S41sh4__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S41sh4.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_이름비석_숲길_b7baf3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "샷 텍스트가 명시한 '얼굴' 중심의 타이트한 구도를 잘 구현했으며, 고기를 내려다보며 찌푸린 표정도 정확합니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "'얼굴'을 강조한 프레이밍 지시와 달리 상반신 전체가 나오는 넓은 샷이며, 명시적으로 없어야 할 펜던트가 가슴에 보입니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "샷 텍스트가 명시한 '얼굴' 중심의 타이트한 구도를 잘 구현했으며, 고기를 내려다보며 찌푸린 표정도 정확합니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "'얼굴'을 강조한 프레이밍 지시와 달리 상반신 전체가 나오는 넓은 샷이며, 명시적으로 없어야 할 펜던트가 가슴에 보입니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "'얼굴'을 강조한 숏 텍스트의 프레이밍과 미간을 찌푸린 표정을 정확히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "프레이밍이 너무 넓어 '얼굴' 중심이 아니며, 지시사항과 달리 목걸이가 노출되었습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "'얼굴'을 강조한 숏 텍스트의 프레이밍과 미간을 찌푸린 표정을 정확히 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "프레이밍이 너무 넓어 '얼굴' 중심이 아니며, 지시사항과 달리 목걸이가 노출되었습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 7
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 15,
   "B": 7
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "샷 텍스트가 명시한 '얼굴' 중심의 타이트한 구도를 잘 구현했으며, 고기를 내려다보며 찌푸린 표정도 정확합니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "'얼굴'을 강조한 프레이밍 지시와 달리 상반신 전체가 나오는 넓은 샷이며, 명시적으로 없어야 할 펜던트가 가슴에 보입니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_이름비석_숲길_b7baf3.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지후가 역겨운 냄새에 미간을 팍 찌푸리는 표정이 아닙니다.",
     "fix_en": "Deepen the furrows between her eyebrows and scrunch her nose slightly to create a strong expression of disgust at a foul smell.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Deepen the furrows between her eyebrows and scrunch her nose slightly to create a strong expression of disgust at a foul smell.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S41sh4__bgfirst_bg.png",
   "bg_asset_id": "701789dd-113d-4674-82ff-d5499a345991",
   "bg_record_key": "S41sh4::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "이름비석_숲길",
   "groupbg_asset_id": "47256cc0-0523-4a95-8f0e-7b08cbca588a"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S41sh7": {
  "input_fingerprint": "6b97070083999408",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 빽빽한 나무 기둥 뒤로 짐승의 털이 난 날카로운 손이 튀어나온 찰나의 클로즈업.\n\nLOCATION (lock): Among tightly spaced pine trunks and undergrowth beside the forest path. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "지시문에 명시된 '짐승의 털이 난 날카로운 손'만을 클로즈업으로 정확히 포착했으며, 다른 인물을 배제하라는 조건을 완벽히 준수했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "털이 난 손의 질감은 표현했으나, 지시문에 없는 인물의 얼굴과 몸 일부가 함께 프레임에 노출되어 감점 요인이 되었습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "손의 묘사가 날카롭지 않으며, 배경에 등장해서는 안 될 인물이 포함되어 '인물 배제' 지시를 명백히 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S41sh4_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "손과 팔이 실제 짐승이나 괴물의 앞발처럼 묘사되어, 모든 인물은 정상적인 해부학적 구조를 가진 인간으로 표현되어야 한다는 규칙을 위반했습니다.",
     "fix_en": "Replace the animal arm and claws with an anatomically normal human arm and hand, featuring human skin and sharp human fingernails, reaching out from behind the tree trunk.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the animal arm and claws with an anatomically normal human arm and hand, featuring human skin and sharp human fingernails, reaching out from behind the tree trunk.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev만 (배경 전용·공유 계획)",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S41sh4"
  },
  "lane_policy": "share_plan_prev_bgonly"
 },
 "S42sh1": {
  "input_fingerprint": "365d4fb620a06c42",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 대낮의 국도변 주유소 주유기 옆에 세워진 구형 아반테 차량의 외부 전경.\n\nLOCATION (lock): At a fuel pump in an open roadside gas station beside a rural highway in daylight. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "레퍼런스의 비석을 정확히 포함하며 구형 아반테가 주유소에 있는 지정된 전경을 충실히 구현함."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "레퍼런스 비석 구조물을 양쪽에 중복으로 배치하는 치명적인 위반이 발생함."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "지정된 레퍼런스 구조물(비석)이 완전히 누락되어 장소 조건을 위반함."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "주유소 캐노피, 기둥, 입간판 및 주유기 등 화면 곳곳에 문자가 포함되어 있습니다 ('No text' 규칙 위반).",
     "fix_en": "Remove all text and typography from the gas station canopy, standing signs, pillars, and pumps, leaving the surfaces blank.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all text and typography from the gas station canopy, standing signs, pillars, and pumps, leaving the surfaces blank.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "seed-bg만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_bypass:bg_only"
 },
 "S42sh4": {
  "input_fingerprint": "3d90012ac41d339d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 정인우의 핸드폰 화면에 여러 섬과 동떨어진 '금월도'라는 글씨가 적힌 지도앱이 선명하게 뜬 찰나.\n\nLOCATION (lock): Beside the compact car at the outdoor roadside fuel pump. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo still wears the cast on his injured arm.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 5,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "신체 노출을 배제한 완벽한 프레이밍과 고립된 '금월도' 지도 묘사를 보여주나, 주유 호스 연속성이 누락됨."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "지시문에 명시되지 않은 손이 등장해 인물 배제(NO PEOPLE) 규칙을 위반함."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "상반신과 팔을 넓게 잡아 인물 배제 규칙을 크게 위반했으며, 차량의 주유 연속성도 틀림."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S42sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 샷 기준 이미지에서 자동차 주유구에 꽂혀 있던 주유 노즐과 주유기로 이어지는 검은색 호스가 사라졌습니다.",
     "fix_en": "Restore the black fuel hose and nozzle connecting the right-side gas pump to the car's fuel door, exactly matching the reference image.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore the black fuel hose and nozzle connecting the right-side gas pump to the car's fuel door, exactly matching the reference image.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev만 (배경 전용·공유 계획)",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S42sh1"
  },
  "lane_policy": "share_plan_prev_bgonly"
 },
 "S43sh5": {
  "input_fingerprint": "19501ca220df0bf3",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 거친 바위 표면에 '수 리 영'이라는 세 글자가 선명하게 음각된 찰나의 클로즈업.\n\nLOCATION (lock): At a mossy stone marker along a looping path deep in the daytime forest. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The same mossy stone and its rotten meat remain unchanged, and the inscription identifies it as a “Su-ri-yeong” grave marker. Jihu’s pendant remains missing.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 5,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "'수 리 영'이라는 세 글자가 바위에 정확하고 선명하게 음각되었으며, 클로즈업 샷의 요구사항을 훌륭히 충족합니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "요구된 텍스트와 환경을 보여주지만 '리' 글자의 형태가 다소 뭉개져 가독성이 떨어집니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "지정된 텍스트의 첫 글자가 '수'가 아닌 '스'에 가깝게 잘못 표기되어 지시사항을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S41sh7_sel.png"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "prev만 (배경 전용·공유 계획)",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S41sh7"
  },
  "lane_policy": "ab_select_bypass:bg_only:share_plan_prev_bgonly"
 },
 "S43sh6": {
  "input_fingerprint": "61066d47ae248ada",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 자신의 이름이 새겨진 비석 앞에서 극심한 공포에 사로잡혀 창백하게 굳어버린 김지후(강수리영)의 얼굴.\n\nLOCATION (lock): Directly before the name-carved marker stone in the dense forest. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the mossy marker stone, rotten meat offering, dense pine forest, and filtered daylight. Exclude the inscription-only framing and show the woman's frightened reaction before it.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The “Su-ri-yeong” inscription and rotten offering remain fixed on the marker; Jihu’s pendant remains missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 5,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "레퍼런스의 비석과 제물 디테일을 완벽히 유지한 채, 극심한 공포에 질려 창백하게 굳은 얼굴을 정확한 클로즈업으로 담아내어 프롬프트를 가장 잘 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "공포에 질린 표정 연기는 훌륭하게 표현되었으나, 기준 이미지와 달리 비석에 새겨진 이름 중 첫 글자('수')가 누락되는 디테일 오류가 발생했습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "비석 앞이라는 위치는 충족했으나, 얼굴 클로즈업이 아닌 다소 넓은 프레이밍을 잡았으며 요구된 '극심한 공포에 사로잡힌' 표정이 너무 밋밋하게 연출되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S43sh5_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "인물의 재킷 소매(화면 좌측 하단)에 텍스트 로고가 있어 '텍스트 금지' 지시를 위반했습니다.",
     "fix_en": "Remove the small white text logo from the jacket sleeve in the lower left corner.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "비석의 '수' 글자 왼쪽, 인물의 뺨 바로 옆에 레퍼런스 이미지에는 없는 정체불명의 글자 조각이 추가되었습니다.",
     "fix_en": "Erase the extra carved lines on the stone immediately to the left of the '수' character, replacing them with bare mossy stone.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the small white text logo from the jacket sleeve in the lower left corner.\n- Erase the extra carved lines on the stone immediately to the left of the '수' character, replacing them with bare mossy stone.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S43sh5"
  }
 },
 "S44sh1": {
  "input_fingerprint": "f28795444381388d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, direct sunlight.\n\nSHOT TEXT (authoritative, Korean): 깨진 창문으로 스며든 햇빛이 핏자국 가득한 도축장 바닥과 고기 걸개를 섬뜩하게 비추는 전경.\n\nLOCATION (lock): Inside the forest slaughterhouse, where midday sunlight enters through a broken window and exposes bloodstains and metal hooks. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Blood from the previous night remains across the slaughterhouse floor and walls, and the slaughter tools remain hanging in place.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "부서진 창문으로 스며드는 햇빛이 핏자국과 고기 걸개를 섬뜩하게 비추는 밀폐된 도축장 내부를 텍스트에 맞게 완벽히 구현했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "창문을 통한 빛은 표현되었으나 양쪽 벽이 개방된 구조로 렌더링되어 텍스트가 의도한 실내의 음산한 분위기가 다소 반감됩니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "부서진 창문이 아닌 개방된 좌측 벽면을 통해 주된 빛이 들어오게 연출하여 텍스트의 핵심 조명 지시를 어겼습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "장소의 구조가 레퍼런스 이미지(숲 속 오솔길의 돌비석)와 전혀 일치하지 않고 완전히 다른 실내 도축장으로 생성되었습니다.",
     "fix_en": "Replace the entire enclosed room environment with the exterior forest path and stone monument structure exactly as shown in the reference photograph.",
     "needs_regeneration": true,
     "unfixable": true
    }
   ]
  },
  "fix_skipped": true,
  "fix_skip_reason": "all_issues_unfixable",
  "ref_mode": "seed-bg만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_bypass:bg_only"
 },
 "S44sh6": {
  "input_fingerprint": "902f67090d4dc94e",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, direct sunlight.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)이 허리를 굽힌 채 엎드린 금월도 마을 청년들 중 한 명의 머리 위에 부드럽게 손을 얹고 있는 찰나.\n\nLOCATION (lock): On the bloodstained slaughterhouse floor before the wall of butchering tools, lit through the broken window. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the bloodstained slaughterhouse floor, hanging tools, broken-window sunlight, and harsh materials. Exclude the empty establishing composition and place the elder over the kneeling youths.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The previous night’s bloodstains and slaughter equipment remain in the room.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 4,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "허리를 굽혀 청년의 머리에 손을 얹는 핵심 동작과 기준 이미지의 배경을 정확하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "기준 배경은 잘 유지했으나, 여성이 청년의 머리가 아닌 등에 손을 얹어 동작 지시를 어겼습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "화면 왼쪽 청년의 머리가 보이지 않고 원근감과 신체 비율이 심하게 왜곡되는 해부학적 오류가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S44sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "가운데 엎드린 청년과 그 왼쪽 뒤에 있는 청년의 몸이 비정상적으로 융합되어 있습니다. 가운데 청년의 하체와 몸통이 뒤쪽 인물과 해부학적으로 불가능하게 엉켜 하나의 덩어리처럼 보입니다.",
     "fix_en": "Separate the middle kneeling youth and the youth behind him into two distinct, anatomically correct human bodies, ensuring both have fully formed torsos and limbs.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Separate the middle kneeling youth and the youth behind him into two distinct, anatomically correct human bodies, ensuring both have fully formed torsos and limbs.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S44sh1"
  }
 },
 "S44sh12": {
  "input_fingerprint": "d4769166427aafec",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, direct sunlight.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)이 붉은 피가 묻은 날카로운 손톱 끝에 혀를 댄 채 기괴한 표정을 짓고 있는 얼굴 클로즈업.\n\nLOCATION (lock): Inside the daylight slaughterhouse beside the collapsed, bleeding youth. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the daylight slaughterhouse look, blood, timber walls, and metallic tools. Exclude the kneeling youths from the close frame and show only the elder licking blood from her claw.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The youth Youngran cut down remains collapsed and bleeding on the slaughterhouse floor among the existing bloodstains.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 8,
   "A": 5,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "날카로운 손톱, 혀를 대는 동작, 기괴한 표정의 클로즈업을 정확히 구현했으며 지시대로 주변 인물을 프레임에서 잘 배제함."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "클로즈업 앵글과 배경은 일치하나, 핵심 디테일인 '날카로운 손톱'이 누락되어 평범한 손가락으로 묘사됨."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "클로즈업 프레임에서 소년들을 배제하라는 명시적인 지시를 어기고 우측 하단에 인물의 일부가 노출됨."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S44sh6_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "여자의 혀가 내밀어져 있지만 프롬프트의 지시대로 피 묻은 날카로운 손톱 끝에 닿지 않았습니다.",
     "fix_en": "Extend the woman's tongue so that its tip makes direct contact with the sharp, bloody tip of the nail on her raised finger.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Extend the woman's tongue so that its tip makes direct contact with the sharp, bloody tip of the nail on her raised finger.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S44sh6"
  }
 },
 "S45sh3::bgfirst_bg": {
  "input_fingerprint": "bb69a4aa08e4cc07",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 우뚝 멈춰 선 채 뒤를 돌아 고산들을 뚫어지게 응시하고 있는 강영란(흰 머리 여자)의 측면.\n\nLOCATION (lock): On a descending daytime forest path outside the hidden slaughterhouse.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height on 강영란(흰 머리 여자)'s lateral side, the track settles into a tightened medium profile with her held at the left foreground and 고산들 retained farther along her sightline at the right. Her abrupt halt and turned head form a rigid horizontal gaze axis across the forest path, compressing the emotional distance without yet moving into the later face-to-face close framing.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the middle-left of the frame, foreground, looks toward 고산들; 고산들 in the middle-right of the frame, background, looks toward 강영란(흰 머리 여자); forest path between them in the lower-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 숲길 (두 인물이 내려오다 멈춘 길); used as The shared ground plane visibly separating the two figures along their locked gaze line; 숲의 나무들 (숲길 양쪽에 자리함) — Tree sides recede behind the left foreground and right background figures; used as Layered vertical framing behind both profiles without interrupting the eyeline.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 우뚝 멈춰 선 채 뒤를 돌아 고산들을 뚫어지게 응시하고 있는 강영란(흰 머리 여자)의 측면.\n\nLOCATION (lock): On a descending daytime forest path outside the hidden slaughterhouse.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height on 강영란(흰 머리 여자)'s lateral side, the track settles into a tightened medium profile with her held at the left foreground and 고산들 retained farther along her sightline at the right. Her abrupt halt and turned head form a rigid horizontal gaze axis across the forest path, compressing the emotional distance without yet moving into the later face-to-face close framing.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the middle-left of the frame, foreground, looks toward 고산들; 고산들 in the middle-right of the frame, background, looks toward 강영란(흰 머리 여자); forest path between them in the lower-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 숲길 (두 인물이 내려오다 멈춘 길); used as The shared ground plane visibly separating the two figures along their locked gaze line; 숲의 나무들 (숲길 양쪽에 자리함) — Tree sides recede behind the left foreground and right background figures; used as Layered vertical framing behind both profiles without interrupting the eyeline.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S45sh3__bgfirst_bg.png",
  "asset_id": "26d7e8b3-b484-4fb3-84d1-b5af770b6b6f",
  "input_asset_ids": [
   "20499f54-fb95-4603-bf38-9b00a162cef2",
   "4240a7d4-8dee-4a7f-9e73-b218ab04ccc6"
  ]
 },
 "S45sh3": {
  "input_fingerprint": "a29230f32f994e1f",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 우뚝 멈춰 선 채 뒤를 돌아 고산들을 뚫어지게 응시하고 있는 강영란(흰 머리 여자)의 측면.\n\nLOCATION (lock): On a descending daytime forest path outside the hidden slaughterhouse. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The youth killed in the slaughterhouse remains there unless removed; Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (육혈귀 각성 상태) (한국인 60대 후반 여성, 고운 노년 얼굴, 틀어 묶은 머리, 길고 날카로운 손톱) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 우뚝 멈춰 선 채 뒤를 돌아 고산들을 뚫어지게 응시하고 있는 강영란(흰 머리 여자)의 측면.\n\nLOCATION (lock): On a descending daytime forest path outside the hidden slaughterhouse. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The youth killed in the slaughterhouse remains there unless removed; Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (육혈귀 각성 상태) (한국인 60대 후반 여성, 고운 노년 얼굴, 틀어 묶은 머리, 길고 날카로운 손톱) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 우뚝 멈춰 선 채 뒤를 돌아 고산들을 뚫어지게 응시하고 있는 강영란(흰 머리 여자)의 측면.\n\nLOCATION (lock): On a descending daytime forest path outside the hidden slaughterhouse. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The youth killed in the slaughterhouse remains there unless removed; Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (육혈귀 각성 상태) (한국인 60대 후반 여성, 고운 노년 얼굴, 틀어 묶은 머리, 길고 날카로운 손톱) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S45sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S45sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자) (육혈귀 각성 상태): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:436856>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L27B02.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자) (육혈귀 각성 상태): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:436856>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "허용되지 않은 인물을 배경에 추가하여 지침을 위반했으며, '뒤를 돌아보는' 동작도 반영되지 않았습니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "뒤를 돌아보는 인물의 측면 모습을 정확히 포착했으며, 레퍼런스의 의상과 분장 디테일도 훌륭하게 재현했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "허용되지 않은 인물을 배경에 추가하여 지침을 위반했으며, '뒤를 돌아보는' 동작도 반영되지 않았습니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "뒤를 돌아보는 인물의 측면 모습을 정확히 포착했으며, 레퍼런스의 의상과 분장 디테일도 훌륭하게 재현했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지문대로 홀로 선 강영란이 뒤돌아보는 측면 모습과 지정된 의상, 날카로운 손톱을 배경에 맞춰 정확히 연출했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "지문에 명시되지 않은 정체불명의 남성을 철조망 너머에 추가로 등장시켜 인물 등장 규칙을 심각하게 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지문대로 홀로 선 강영란이 뒤돌아보는 측면 모습과 지정된 의상, 날카로운 손톱을 배경에 맞춰 정확히 연출했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지문에 명시되지 않은 정체불명의 남성을 철조망 너머에 추가로 등장시켜 인물 등장 규칙을 심각하게 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "허용되지 않은 인물을 배경에 추가하여 지침을 위반했으며, '뒤를 돌아보는' 동작도 반영되지 않았습니다."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "뒤를 돌아보는 인물의 측면 모습을 정확히 포착했으며, 레퍼런스의 의상과 분장 디테일도 훌륭하게 재현했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L27B02.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자) (육혈귀 각성 상태): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:436856>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "왼쪽 손의 해부학적 구조가 뭉개져 있고 손가락 관절이 기형적입니다.",
     "fix_en": "Repair the left hand to have correct human anatomy with five distinct fingers, while keeping the long sharp nails.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Repair the left hand to have correct human anatomy with five distinct fingers, while keeping the long sharp nails.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S45sh3__bgfirst_bg.png",
   "bg_asset_id": "26d7e8b3-b484-4fb3-84d1-b5af770b6b6f",
   "bg_record_key": "S45sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S45sh8": {
  "input_fingerprint": "f402939835dd2451",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 흙바닥에 머리를 댄 채 절망적인 표정으로 눈물을 머금고 엎드려 있는 고산들의 모습.\n\nLOCATION (lock): On the dirt of the forest path where the two have stopped among the trees. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the daytime pine path, earthen ground, filtered light, and the two walkers' clothing. Exclude their earlier standing separation and show the man prostrated before the elder.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 6,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "짧게 깎은 머리와 흙바닥에 엎드려 눈물 짓는 모습을 잘 구현했으며, 샷 텍스트에 없는 인물을 배제하는 제한 규칙을 정확히 준수했습니다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "자세와 표정은 지시문과 일치하나, 샷 텍스트에 등장하지 않는 다른 인물(노인)의 일부가 프레임에 포함되어 엄격한 인물 제한 규칙을 위반했습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "절망하며 엎드린 묘사는 있으나, '짧게 깎은 검은 머리'라는 명확한 캐릭터 외양 설정과 달리 머리카락이 길게 묘사되어 감점되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S45sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "캐릭터 레퍼런스 이미지에서 착용하고 있는 검은색 비니가 누락되었습니다.",
     "fix_en": "Add the black knit beanie from the character reference onto the man's head.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the black knit beanie from the character reference onto the man's head.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S45sh3"
  }
 },
 "S45sh10": {
  "input_fingerprint": "96373206e813e397",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 고산들을 꼭 끌어안은 강영란(흰 머리 여자)이 비장한 눈빛으로 허공을 노려보는 굳은 얼굴 클로즈업.\n\nLOCATION (lock): On the secluded forest path, where both figures kneel on the ground beneath the trees. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the forest path, daylight, and the kneeling pair's positions. Exclude the earlier prostration and show the elder embracing the man.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (육혈귀 각성 상태) (한국인 60대 후반 여성, 고운 노년 얼굴, 틀어 묶은 머리, 길고 날카로운 손톱); 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 2,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "치명적인 해부학적 오류 없이 두 인물의 포옹과 여성의 비장한 표정을 잘 담아냈으나, 이전 샷에서 이어지는 남성 옷의 진흙 디테일이 다소 부족합니다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "남성 등에 얹힌 여성의 손 구조(엄지 방향)가 물리적으로 불가능하고 하단에 출처 모를 손가락이 나타나는 등 심각한 신체 오류가 있습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "클로즈업 샷의 구도와 허공을 노려보는 굳은 표정, 옷의 흙먼지 묘사는 가장 완벽하지만, 여성의 손가락이 6개로 렌더링된 치명적인 단점이 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S45sh8_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자) (육혈귀 각성 상태): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:436856>"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "강영란이 레퍼런스의 흰색 한복이 아닌 검은색 옷을 입고 있습니다.",
     "fix_en": "Change the woman's black clothing to a white hanbok top, matching her character reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the woman's black clothing to a white hanbok top, matching her character reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S45sh8"
  }
 },
 "S46sh3::bgfirst_bg": {
  "input_fingerprint": "e1cdc31aed1fd621",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 자신의 이름이 새겨진 비석 앞에서 극심한 공포에 질려 눈을 크게 뜬 채 굳어 있는 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): Among a cluster of name-carved marker stones along the daytime forest route.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: The advancing camera finishes beside the grave marker in an extreme facial close-up of 김지후(강수리영), with her widened eyes and arrested breath occupying most of the right side. A narrow edge of the marker remains at the left border with the carved name legible enough to bind her terror to the place.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 김지후(강수리영) in the middle-right of the frame, foreground; 수리영 이름이 새겨진 비석 가장자리 in the middle-left of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 비석 (‘수리영’이라는 이름이 새겨져 있음) — The inscribed face is angled toward camera, showing the name ‘수리영’ along the visible edge; used as Its narrow edge provides spatial context beside the extreme close-up; 숲속 (낮의 숲속); used as The surrounding forest remains only as soft peripheral context.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 자신의 이름이 새겨진 비석 앞에서 극심한 공포에 질려 눈을 크게 뜬 채 굳어 있는 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): Among a cluster of name-carved marker stones along the daytime forest route.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: The advancing camera finishes beside the grave marker in an extreme facial close-up of 김지후(강수리영), with her widened eyes and arrested breath occupying most of the right side. A narrow edge of the marker remains at the left border with the carved name legible enough to bind her terror to the place.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 김지후(강수리영) in the middle-right of the frame, foreground; 수리영 이름이 새겨진 비석 가장자리 in the middle-left of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 비석 (‘수리영’이라는 이름이 새겨져 있음) — The inscribed face is angled toward camera, showing the name ‘수리영’ along the visible edge; used as Its narrow edge provides spatial context beside the extreme close-up; 숲속 (낮의 숲속); used as The surrounding forest remains only as soft peripheral context.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S46sh3__bgfirst_bg.png",
  "asset_id": "bb1661c3-0a46-44c5-ac69-cf459b0a7511",
  "input_asset_ids": [
   "801b6302-f95c-4a29-8459-76830cef77e0",
   "47256cc0-0523-4a95-8f0e-7b08cbca588a"
  ]
 },
 "S46sh3": {
  "input_fingerprint": "9359a9e8da6bffe5",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 자신의 이름이 새겨진 비석 앞에서 극심한 공포에 질려 눈을 크게 뜬 채 굳어 있는 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): Among a cluster of name-carved marker stones along the daytime forest route. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Multiple mossy stones remain inscribed “Su-ri-yeong,” each with rotten meat before it. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 자신의 이름이 새겨진 비석 앞에서 극심한 공포에 질려 눈을 크게 뜬 채 굳어 있는 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): Among a cluster of name-carved marker stones along the daytime forest route. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Multiple mossy stones remain inscribed “Su-ri-yeong,” each with rotten meat before it. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 자신의 이름이 새겨진 비석 앞에서 극심한 공포에 질려 눈을 크게 뜬 채 굳어 있는 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): Among a cluster of name-carved marker stones along the daytime forest route. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Multiple mossy stones remain inscribed “Su-ri-yeong,” each with rotten meat before it. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S46sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S46sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1226689>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_이름비석_숲길_b7baf3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1226689>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "요구된 '얼굴 클로즈업' 프레이밍과 극심한 공포 표정을 정확히 연출했으며, 지정된 인물 레퍼런스를 충실히 반영했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "'얼굴 클로즈업' 지시를 어기고 프레임을 넓혔으며, 지정된 인물(고산들)이 아닌 임의의 여성을 생성하여 규정을 위반했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "요구된 '얼굴 클로즈업' 프레이밍과 극심한 공포 표정을 정확히 연출했으며, 지정된 인물 레퍼런스를 충실히 반영했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "'얼굴 클로즈업' 지시를 어기고 프레임을 넓혔으며, 지정된 인물(고산들)이 아닌 임의의 여성을 생성하여 규정을 위반했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 캐릭터(고산들) 레퍼런스를 정확히 반영하였으며, 요구된 얼굴 클로즈업과 극심한 공포에 질린 표정을 훌륭하게 연출했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "허용되지 않은 여성 인물이 등장하여 인물 제약을 위반했으며, 프레이밍이 얼굴 클로즈업보다 너무 넓게 잡혔습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 캐릭터(고산들) 레퍼런스를 정확히 반영하였으며, 요구된 얼굴 클로즈업과 극심한 공포에 질린 표정을 훌륭하게 연출했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "허용되지 않은 여성 인물이 등장하여 인물 제약을 위반했으며, 프레이밍이 얼굴 클로즈업보다 너무 넓게 잡혔습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 6
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "요구된 '얼굴 클로즈업' 프레이밍과 극심한 공포 표정을 정확히 연출했으며, 지정된 인물 레퍼런스를 충실히 반영했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "'얼굴 클로즈업' 지시를 어기고 프레임을 넓혔으며, 지정된 인물(고산들)이 아닌 임의의 여성을 생성하여 규정을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_이름비석_숲길_b7baf3.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1226689>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "인물의 얼굴 크기가 레이아웃 스케치의 타이트한 클로즈업 비율보다 훨씬 작게 렌더링되어 이마와 머리 윗부분이 프레임 안에 다 들어와 있습니다.",
     "fix_en": "Scale up the person's head and face significantly to match the tight close-up size of the layout sketch, ensuring the forehead extends beyond the top edge of the frame.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Scale up the person's head and face significantly to match the tight close-up size of the layout sketch, ensuring the forehead extends beyond the top edge of the frame.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S46sh3__bgfirst_bg.png",
   "bg_asset_id": "bb1661c3-0a46-44c5-ac69-cf459b0a7511",
   "bg_record_key": "S46sh3::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "이름비석_숲길",
   "groupbg_asset_id": "47256cc0-0523-4a95-8f0e-7b08cbca588a"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S46sh11": {
  "input_fingerprint": "0def665e74f8b9ec",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 눈을 크게 뜬 김지후(강수리영)의 시선 끝에 우뚝 서 있는 고산들의 듬직한 전신.\n\nLOCATION (lock): In a forest clearing beside the path, between the fallen woman and the injured black goat. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the mossy grave markers, dense pine trunks, undergrowth, and filtered daylight. Exclude the earlier isolated frightened close-up and place the man at the woman's sightline.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The huge black goat lies incapacitated in the forest, breathing only shallowly. Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 5,
   "A": 2
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "여자의 '크게 뜬 눈'과 남성의 전신을 프레임에 정확히 담아내 텍스트 구도를 잘 살렸으나, 남성의 바지 색상(우선순위 4)이 레퍼런스와 다릅니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "남성의 복장은 정확히 구현했지만, 여자를 뒷모습으로 배치하여 샷 텍스트가 요구한 '눈을 크게 뜬' 연출(우선순위 2)을 보여주지 못했습니다."
   },
   {
    "label": "A",
    "score": 2,
    "verdict_ko": "이전 샷의 클로즈업 인물을 제외하라는 명시적 지시를 무시하고, 허용되지 않은 인물을 전면에 배치하여 심각한 하드 위반(우선순위 1)을 범했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S46sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "고산들의 의상이 이전 샷에 설정된 네이비 정장과 흰 셔츠가 아닌 전술 재킷과 바지로 변경되었습니다.",
     "fix_en": "Change the man's clothing to a navy blue suit jacket and a white collared shirt to match the previous shot.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "샷 텍스트에 남성의 '전신'이 명시되어 있으나 프레임 하단이 잘려 전체 모습이 보이지 않습니다.",
     "fix_en": "Widen the camera framing to include the man's feet and show his complete full body.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "무력화되어 쓰러져 있어야 할 검은 염소가 머리를 꼿꼿이 들고 앉아 있습니다.",
     "fix_en": "Modify the goat's pose to lie completely flat on the ground with its head down.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the man's clothing to a navy blue suit jacket and a white collared shirt to match the previous shot.\n- Modify the goat's pose to lie completely flat on the ground with its head down.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S46sh3"
  }
 },
 "S46sh16": {
  "input_fingerprint": "e45f1dcc97cef97b",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 고산들이 쓰러진 수리영을 공격한 대형 검은 염소의 커다란 몸통을 어깨 위로 훌쩍 둘러멘 듬직한 뒷모습.\n\nLOCATION (lock): On the descending woodland path beside the spot where the large black goat fell. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same dense forest, mossy stones, earthen path, and daylight. Exclude the earlier confrontation and show the man carrying the fallen black goat away.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul carries the incapacitated black goat over his shoulder; Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 5,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "이전 컷에서 확립된 의상(파란 재킷)과 배경(비석)을 정확히 유지하면서, 카메라 구도를 자연스럽게 변경해 염소를 짊어진 뒷모습을 성공적으로 구현했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "염소를 짊어진 뒷모습은 잘 표현되었으나, 이전 컷과 인물 의상이 다르고 장소의 핵심 요소인 비석이 프레임에서 누락되었습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "이전 컷의 카메라 구도와 쓰러진 인물의 위치를 그대로 복사하여 '카메라 프레이밍을 복사하지 말 것'이라는 지침을 명백히 위반했으며, 의상도 유지되지 않았습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S46sh11_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 장면에 있던 글자가 새겨진 길쭉한 비석이 글자가 없는 작고 둥근 비석으로 바뀌었습니다.",
     "fix_en": "Replace the small, rounded, blank stone on the left with the tall, flat stone monument with carved characters from the previous shot still.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the small, rounded, blank stone on the left with the tall, flat stone monument with carved characters from the previous shot still.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S46sh11"
  }
 },
 "S47sh1::bgfirst_bg": {
  "input_fingerprint": "5b92c819b5c38649",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 대낮의 해안가 선착장에서 드럼통 모닥불 곁에 쪼그리고 앉은 정인우가 수심이 가득한 표정으로 낚싯꾼 #1을 올려다보는 모습.\n\nLOCATION (lock): Beside a barrel fire on a mainland coastal pier in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At crouching eye height beside the group, a static wide composition places 정인우 low in the left foreground, folded beside the barrel fire, while 낚싯꾼 #1 sits across from him in the right midground. The diagonal between 정인우's lifted face and the fisherman's indifferent posture keeps the busy pier secondary but legible behind them.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 드럼통 모닥불 (불이 피워져 있음) — The barrel's side and upper opening are visible obliquely between 정인우 and the fishermen; used as The burning barrel sits between the conversational planes and provides a visual center for the group; 소주잔 (어부들이 술을 마시는 중) — Their open rims and sides are visible beside the fishermen's hands; used as Glasses remain near the seated fishermen as evidence of their unconcerned drinking; 선착장 (낮의 해안마을 선착장) — Its length recedes behind the seated group toward the far end; used as The pier extends behind the group and establishes the coastal working location.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 대낮의 해안가 선착장에서 드럼통 모닥불 곁에 쪼그리고 앉은 정인우가 수심이 가득한 표정으로 낚싯꾼 #1을 올려다보는 모습.\n\nLOCATION (lock): Beside a barrel fire on a mainland coastal pier in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At crouching eye height beside the group, a static wide composition places 정인우 low in the left foreground, folded beside the barrel fire, while 낚싯꾼 #1 sits across from him in the right midground. The diagonal between 정인우's lifted face and the fisherman's indifferent posture keeps the busy pier secondary but legible behind them.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 드럼통 모닥불 (불이 피워져 있음) — The barrel's side and upper opening are visible obliquely between 정인우 and the fishermen; used as The burning barrel sits between the conversational planes and provides a visual center for the group; 소주잔 (어부들이 술을 마시는 중) — Their open rims and sides are visible beside the fishermen's hands; used as Glasses remain near the seated fishermen as evidence of their unconcerned drinking; 선착장 (낮의 해안마을 선착장) — Its length recedes behind the seated group toward the far end; used as The pier extends behind the group and establishes the coastal working location.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S47sh1__bgfirst_bg.png",
  "asset_id": "dc5514bf-85cd-4905-ac4e-b532eeaf402b",
  "input_asset_ids": [
   "ac5c9172-0412-4e3c-a168-27292f6e2384",
   "803cc956-9eff-49c4-b977-3e62ad55fc55"
  ]
 },
 "S47sh1": {
  "input_fingerprint": "daa09bab06a588ea",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 대낮의 해안가 선착장에서 드럼통 모닥불 곁에 쪼그리고 앉은 정인우가 수심이 가득한 표정으로 낚싯꾼 #1을 올려다보는 모습.\n\nLOCATION (lock): Beside a barrel fire on a mainland coastal pier in daylight. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 대낮의 해안가 선착장에서 드럼통 모닥불 곁에 쪼그리고 앉은 정인우가 수심이 가득한 표정으로 낚싯꾼 #1을 올려다보는 모습.\n\nLOCATION (lock): Beside a barrel fire on a mainland coastal pier in daylight. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 대낮의 해안가 선착장에서 드럼통 모닥불 곁에 쪼그리고 앉은 정인우가 수심이 가득한 표정으로 낚싯꾼 #1을 올려다보는 모습.\n\nLOCATION (lock): Beside a barrel fire on a mainland coastal pier in daylight. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S47sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S47sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "위치 레퍼런스(비석과 산책로)를 정확히 배경에 반영하였으며, 인물의 쪼그린 자세, 깁스 설정, 수심 가득한 표정 등 샷 텍스트의 지시를 충실히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "화면에 검은색 지시선(화살표) 마커가 그대로 노출되는 치명적인 오류(Hard Violation)가 발생했으며, 필수 위치 레퍼런스 구조물도 누락되었습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "위치 레퍼런스(비석과 산책로)를 정확히 배경에 반영하였으며, 인물의 쪼그린 자세, 깁스 설정, 수심 가득한 표정 등 샷 텍스트의 지시를 충실히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "화면에 검은색 지시선(화살표) 마커가 그대로 노출되는 치명적인 오류(Hard Violation)가 발생했으며, 필수 위치 레퍼런스 구조물도 누락되었습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지정된 장소(비석)와 시간대, 인물의 자세와 깁스 상태, 낚시꾼을 올려다보는 표정까지 프롬프트의 지시사항을 매우 충실히 반영했습니다."
     },
     {
      "label": "B",
      "score": 2,
      "verdict_ko": "이미지 상에 검은 화살표(마커)가 노출되는 치명적인 규정 위반(Hard Violation)이 발생했으며, 레퍼런스의 장소(비석)도 누락되었습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 장소(비석)와 시간대, 인물의 자세와 깁스 상태, 낚시꾼을 올려다보는 표정까지 프롬프트의 지시사항을 매우 충실히 반영했습니다."
     },
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "이미지 상에 검은 화살표(마커)가 노출되는 치명적인 규정 위반(Hard Violation)이 발생했으며, 레퍼런스의 장소(비석)도 누락되었습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 5,
     "B": 15
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 5,
   "B": 15
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "위치 레퍼런스(비석과 산책로)를 정확히 배경에 반영하였으며, 인물의 쪼그린 자세, 깁스 설정, 수심 가득한 표정 등 샷 텍스트의 지시를 충실히 구현했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "화면에 검은색 지시선(화살표) 마커가 그대로 노출되는 치명적인 오류(Hard Violation)가 발생했으며, 필수 위치 레퍼런스 구조물도 누락되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "낚시꾼이 쥐고 있는 낚싯대 부분에서 화면 왼쪽으로 굵은 막대가 수평으로 잘못 뻗어 나와 있습니다.",
     "fix_en": "Remove the thick horizontal pole extending to the left from the fisherman's hands, keeping only the single rod pointing diagonally upwards.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "릴 손잡이를 돌리는 낚시꾼의 오른손 손가락이 뭉개져 형태를 알아볼 수 없습니다.",
     "fix_en": "Render clear, anatomically correct fingers on the fisherman's right hand gripping the reel handle.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the thick horizontal pole extending to the left from the fisherman's hands, keeping only the single rod pointing diagonally upwards.\n- Render clear, anatomically correct fingers on the fisherman's right hand gripping the reel handle.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S47sh1__bgfirst_bg.png",
   "bg_asset_id": "dc5514bf-85cd-4905-ac4e-b532eeaf402b",
   "bg_record_key": "S47sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S47sh7": {
  "input_fingerprint": "f01a4107d055f8ae",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 폴리스 라인 안쪽 바닥에 머리와 팔 한쪽이 참혹하게 뜯겨나간 시신이 핏물과 함께 널브러진 끔찍한 상태.\n\nLOCATION (lock): Inside a police cordon at the far end of the mainland pier, around a mutilated body on the ground. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The unidentified corpse is sprawled flat on the ground inside the police line, with the head and one entire arm missing and the remaining torso and limbs resting in pooled blood.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 4,
   "B": 3,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "레퍼런스의 배경과 폴리스 라인은 훌륭하게 재현했으나, 핵심인 훼손된 시신과 핏물이 텐트로 가려져 보이지 않습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "시신이 가림막으로 대체되어 주요 요구 사항이 누락되었고, 원본에 없는 철제 난간이 추가되었습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "시신과 혈흔 묘사가 완전히 누락되었으며, 비석 등 배경 구조물의 디테일도 레퍼런스와 차이가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S47sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "머리와 팔이 훼손된 채 핏물에 널브러진 시신 대신 검은 텐트가 중앙을 가리고 있습니다.",
     "fix_en": "Replace the central black tent with a sprawled corpse missing its head and one arm, resting in pooled blood.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "레퍼런스의 맑은 날씨와 달리 배경에 짙은 안개가 끼어 있습니다.",
     "fix_en": "Remove the fog and restore clear daylight visibility in the background to match the reference.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "레퍼런스의 부두에는 존재하지 않는 철제 난간이 양쪽 가장자리에 추가되었습니다.",
     "fix_en": "Remove the rusty metal railings running along the left and right edges of the pier.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "프롬프트의 금지 지시와 달리 폴리스 라인 테이프와 증거 표지판에 텍스트와 숫자가 있습니다.",
     "fix_en": "Remove all text and numbers from the yellow police tapes and the evidence markers on the ground.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the central black tent with a sprawled corpse missing its head and one arm, resting in pooled blood.\n- Remove the fog and restore clear daylight visibility in the background to match the reference.\n- Remove the rusty metal railings running along the left and right edges of the pier.\n- Remove all text and numbers from the yellow police tapes and the evidence markers on the ground.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev만 (배경 전용·공유 계획)",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S47sh1"
  },
  "lane_policy": "share_plan_prev_bgonly"
 },
 "S47sh10::bgfirst_bg": {
  "input_fingerprint": "076acb92489cda57",
  "prompt": "Turn the attached storyboard sketch into a photorealistic film still of\nits LOCATION ONLY, keeping the figures exactly as they are.\n\nKEEP EXACTLY: the camera framing, the horizon, and where every element\nsits in the frame. Each mannequin figure stays a plain grey featureless\nmannequin standing in the very same spot, at the same size, in the same\npose, turned the same way — do not turn them into people, do not move,\nrotate, mirror or re-pose them, do not add or remove figures.\n\nBUILD PHOTOREALISTICALLY: everything that is not a figure — ground,\nsurfacing, structures, vegetation, sky, water, distance. The bare lines\nof the sketch are a layout guide; replace them with the real materials,\ndepth and lighting of the place described below.\n\nSHOT TEXT this background must serve (Korean): 바닥에서 몸을 일으키던 조 형사가 폴리스 라인 너머 인파 속의 정인우와 똑바로 눈이 마주친 채 멈칫하는 순간.\n\nLOCATION (lock): At the edge of the pier crime scene, between the body inside the cordon and the onlooking crowd beyond it.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At standing eye height beside the investigation area, a slightly tightened side-on composition places 조 형사 at left midground, caught halfway upright, and 정인우 at right background between members of the crowd. The police line cuts horizontally between their depth planes as their opposing profiles lock into a single uninterrupted sightline.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 조 형사 in the middle-left of the frame, midground, looks toward 정인우 beyond the police line; 정인우 in the middle-right of the frame, background, looks toward 조 형사 inside the investigation area; police line dividing the two men in the lower-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 폴리스 라인 (경찰이 접근을 막기 위해 설치함) — It crosses laterally between 조 형사 inside and 정인우 outside the restricted area; used as The barrier remains visibly between the detective and 정인우, turning their recognition into a divided two-plane composition; 구경꾼 인파 (폴리스 라인 밖에 몰려 있음); used as The crowd partly encloses 정인우 while leaving his face visible along the shared eyeline; 구경꾼들의 핸드폰 (사진과 셀카를 찍는 데 사용 중) — Several screen faces are directed toward the investigation area as people photograph the scene; used as Raised phones punctuate the background around 정인우 without blocking the two men's gaze.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nTHINGS AT THIS PLACE:\nThe site is organized along a forest path that enters from the outer forest zone and proceeds inward through a dense stand of pine trunks. At its inner end, the recognizable path disappears into a trackless deep-forest interior. Beyond and around this transition, several stone markers inscribed with Suri-young's name are distributed at separate points along the continuing walking line. Dense pine trunks enclose the route on both sides and continue past the path disappearance point. Undergrowth fills the spaces between the trunks, borders the path, and occupies the off-path ground around the dispersed marker stones.\n- path: Outer-to-deep forest path\n- path end: Forest path disappearance point\n- marker stone: Group of stone markers inscribed with Suri-young's name\n- tree stand: Dense pine trunk stand\n- undergrowth: Intervening forest undergrowth\n\nWORLD FACTS (creator-confirmed — always true):\n- Region (real-world reference): 대한민국 서울권과 서남해안 일대, 대한민국 영토 내 서해의 가상 외딴섬 공동체\n- Era: 주 배경은 스마트폰과 현대 경찰·교통 체계가 존재하는 2020년대 중반의 현대이며, 1670년 조선시대와 20세기 전반 일제강점기 전후의 역사적 재현 구간이 명확히 구분되어 병존한다.\n- 현대 구간은 2020년대 대한민국의 생활상으로, 1670년대 구간은 조선 후기의 복식과 생활 도구로, 20세기 전반 구간은 당시 조선인 복식과 일본군 장비로 표현한다. 시대 전환이 없는 프레임에는 서로 다른 시대의 의상·무기·교통수단·전자기기를 혼합하지 않는다.\n- 현대 육지에서는 스마트폰, 메신저 앱, 지도 앱, CCTV, 바코드 출근 시스템, 현대 경찰 장비, 승용차·구형 SUV·동력 어선이 자연스럽게 존재한다. 섬에서는 휴대전화 신호와 인터넷 연결이 불안정하거나 차단되며, 제한적인 전기 조명과 오래된 생활 설비를 우선하되 현대 물품 자체가 전혀 없는 원시사회로 오해하지 않는다. 역사 구간에는 디지털 기기와 현대 차량을 배제한다.\n\nTHINGS THAT LIVE AT THIS PLACE (what they are, not where they go): the\nlist above names what permanently belongs to this site. Where each one\nsits in this frame is decided by the attached sketch alone — do not add\nor reposition anything on the strength of the list, build only what the\nsketch already has lines for, and do not invent facilities that are not\nlisted.\n\nNo text, no captions, no diagram symbols, no markers anywhere in the\nimage.",
  "effective_prompt": "Turn the attached storyboard sketch into a photorealistic film still of\nits LOCATION ONLY, keeping the figures exactly as they are.\n\nKEEP EXACTLY: the camera framing, the horizon, and where every element\nsits in the frame. Each mannequin figure stays a plain grey featureless\nmannequin standing in the very same spot, at the same size, in the same\npose, turned the same way — do not turn them into people, do not move,\nrotate, mirror or re-pose them, do not add or remove figures.\n\nBUILD PHOTOREALISTICALLY: everything that is not a figure — ground,\nsurfacing, structures, vegetation, sky, water, distance. The bare lines\nof the sketch are a layout guide; replace them with the real materials,\ndepth and lighting of the place described below.\n\nSHOT TEXT this background must serve (Korean): 바닥에서 몸을 일으키던 조 형사가 폴리스 라인 너머 인파 속의 정인우와 똑바로 눈이 마주친 채 멈칫하는 순간.\n\nLOCATION (lock): At the edge of the pier crime scene, between the body inside the cordon and the onlooking crowd beyond it.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At standing eye height beside the investigation area, a slightly tightened side-on composition places 조 형사 at left midground, caught halfway upright, and 정인우 at right background between members of the crowd. The police line cuts horizontally between their depth planes as their opposing profiles lock into a single uninterrupted sightline.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 조 형사 in the middle-left of the frame, midground, looks toward 정인우 beyond the police line; 정인우 in the middle-right of the frame, background, looks toward 조 형사 inside the investigation area; police line dividing the two men in the lower-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 폴리스 라인 (경찰이 접근을 막기 위해 설치함) — It crosses laterally between 조 형사 inside and 정인우 outside the restricted area; used as The barrier remains visibly between the detective and 정인우, turning their recognition into a divided two-plane composition; 구경꾼 인파 (폴리스 라인 밖에 몰려 있음); used as The crowd partly encloses 정인우 while leaving his face visible along the shared eyeline; 구경꾼들의 핸드폰 (사진과 셀카를 찍는 데 사용 중) — Several screen faces are directed toward the investigation area as people photograph the scene; used as Raised phones punctuate the background around 정인우 without blocking the two men's gaze.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nTHINGS AT THIS PLACE:\nThe site is organized along a forest path that enters from the outer forest zone and proceeds inward through a dense stand of pine trunks. At its inner end, the recognizable path disappears into a trackless deep-forest interior. Beyond and around this transition, several stone markers inscribed with Suri-young's name are distributed at separate points along the continuing walking line. Dense pine trunks enclose the route on both sides and continue past the path disappearance point. Undergrowth fills the spaces between the trunks, borders the path, and occupies the off-path ground around the dispersed marker stones.\n- path: Outer-to-deep forest path\n- path end: Forest path disappearance point\n- marker stone: Group of stone markers inscribed with Suri-young's name\n- tree stand: Dense pine trunk stand\n- undergrowth: Intervening forest undergrowth\n\nWORLD FACTS (creator-confirmed — always true):\n- Region (real-world reference): 대한민국 서울권과 서남해안 일대, 대한민국 영토 내 서해의 가상 외딴섬 공동체\n- Era: 주 배경은 스마트폰과 현대 경찰·교통 체계가 존재하는 2020년대 중반의 현대이며, 1670년 조선시대와 20세기 전반 일제강점기 전후의 역사적 재현 구간이 명확히 구분되어 병존한다.\n- 현대 구간은 2020년대 대한민국의 생활상으로, 1670년대 구간은 조선 후기의 복식과 생활 도구로, 20세기 전반 구간은 당시 조선인 복식과 일본군 장비로 표현한다. 시대 전환이 없는 프레임에는 서로 다른 시대의 의상·무기·교통수단·전자기기를 혼합하지 않는다.\n- 현대 육지에서는 스마트폰, 메신저 앱, 지도 앱, CCTV, 바코드 출근 시스템, 현대 경찰 장비, 승용차·구형 SUV·동력 어선이 자연스럽게 존재한다. 섬에서는 휴대전화 신호와 인터넷 연결이 불안정하거나 차단되며, 제한적인 전기 조명과 오래된 생활 설비를 우선하되 현대 물품 자체가 전혀 없는 원시사회로 오해하지 않는다. 역사 구간에는 디지털 기기와 현대 차량을 배제한다.\n\nTHINGS THAT LIVE AT THIS PLACE (what they are, not where they go): the\nlist above names what permanently belongs to this site. Where each one\nsits in this frame is decided by the attached sketch alone — do not add\nor reposition anything on the strength of the list, build only what the\nsketch already has lines for, and do not invent facilities that are not\nlisted.\n\nNo text, no captions, no diagram symbols, no markers anywhere in the\nimage.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S47sh10__bgfirst_bg.png",
  "asset_id": "b667c132-a700-48ba-aaeb-f5f3496a2817",
  "input_asset_ids": [
   "c4b76061-f014-4d6e-aa45-a1cbcc2bb6f0"
  ]
 },
 "S47sh10": {
  "input_fingerprint": "ec7b8a09cb15edaa",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 바닥에서 몸을 일으키던 조 형사가 폴리스 라인 너머 인파 속의 정인우와 똑바로 눈이 마주친 채 멈칫하는 순간.\n\nLOCATION (lock): At the edge of the pier crime scene, between the body inside the cordon and the onlooking crowd beyond it. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the crowded harbor crime scene, police tape, mutilated body, and daylight. Exclude the body-only emphasis and show the detective looking up toward the man in the crowd.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The mutilated corpse remains on the ground inside the police perimeter; Inwoo still wears his arm cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Replace every grey or outlined mannequin figure in the FIRST attached\nimage with the real people the CHARACTER REFERENCE images show, and\noutput a photorealistic film still. Each mannequin becomes the character\nwhose POSE CANON and the shot text match that figure's pose and\nposition; never repeat one person across two mannequins. If no CHARACTER\nREFERENCE image is attached for a figure, still replace it with a\nplausible real person consistent with the text. Every mannequin becomes\na person — erasing a figure, or leaving one out, is not a replacement,\nand the number of figures never changes.\n\nKEEP EXACTLY: the background of the first image, the camera framing,\nand each mannequin's position, size, pose and the direction it is\nturned. A person must be exactly where their mannequin was, at the same\nscale, in the same pose, facing the same way — never mirrored, never\nre-staged, never straightened up, stood upright or otherwise re-posed.\nNo mannequin, grey figure or sketch line may remain anywhere in the\noutput.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 바닥에서 몸을 일으키던 조 형사가 폴리스 라인 너머 인파 속의 정인우와 똑바로 눈이 마주친 채 멈칫하는 순간.\n\nLOCATION (lock): At the edge of the pier crime scene, between the body inside the cordon and the onlooking crowd beyond it. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The mutilated corpse remains on the ground inside the police perimeter; Inwoo still wears his arm cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Replace every grey or outlined mannequin figure in the FIRST attached\nimage with the real people the CHARACTER REFERENCE images show, and\noutput a photorealistic film still. Each mannequin becomes the character\nwhose POSE CANON and the shot text match that figure's pose and\nposition; never repeat one person across two mannequins. If no CHARACTER\nREFERENCE image is attached for a figure, still replace it with a\nplausible real person consistent with the text. Every mannequin becomes\na person — erasing a figure, or leaving one out, is not a replacement,\nand the number of figures never changes.\n\nKEEP EXACTLY: the background of the first image, the camera framing,\nand each mannequin's position, size, pose and the direction it is\nturned. A person must be exactly where their mannequin was, at the same\nscale, in the same pose, facing the same way — never mirrored, never\nre-staged, never straightened up, stood upright or otherwise re-posed.\nNo mannequin, grey figure or sketch line may remain anywhere in the\noutput.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 바닥에서 몸을 일으키던 조 형사가 폴리스 라인 너머 인파 속의 정인우와 똑바로 눈이 마주친 채 멈칫하는 순간.\n\nLOCATION (lock): At the edge of the pier crime scene, between the body inside the cordon and the onlooking crowd beyond it. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The mutilated corpse remains on the ground inside the police perimeter; Inwoo still wears his arm cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "C": "Replace every grey or outlined mannequin figure in the FIRST attached\nimage with the real people the CHARACTER REFERENCE images show, and\noutput a photorealistic film still. Each mannequin becomes the character\nwhose POSE CANON and the shot text match that figure's pose and\nposition; never repeat one person across two mannequins. If no CHARACTER\nREFERENCE image is attached for a figure, still replace it with a\nplausible real person consistent with the text. Every mannequin becomes\na person — erasing a figure, or leaving one out, is not a replacement,\nand the number of figures never changes.\n\nKEEP EXACTLY: the background of the first image, the camera framing,\nand each mannequin's position, size, pose and the direction it is\nturned. A person must be exactly where their mannequin was, at the same\nscale, in the same pose, facing the same way — never mirrored, never\nre-staged, never straightened up, stood upright or otherwise re-posed.\nNo mannequin, grey figure or sketch line may remain anywhere in the\noutput.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 바닥에서 몸을 일으키던 조 형사가 폴리스 라인 너머 인파 속의 정인우와 똑바로 눈이 마주친 채 멈칫하는 순간.\n\nLOCATION (lock): At the edge of the pier crime scene, between the body inside the cordon and the onlooking crowd beyond it. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The mutilated corpse remains on the ground inside the police perimeter; Inwoo still wears his arm cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S47sh10__bgfirst_bg.png"
    }
   ],
   "B": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S47sh10__bgfirst_bg.png"
    }
   ],
   "C": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S47sh10__bgfirst_bg.png"
    }
   ]
  },
  "totals": {
   "A": 7,
   "B": 6,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "레퍼런스의 배경과 구도를 정확히 유지하며, 몸을 일으키는 형사와 깁스를 한 정인우의 시선 교환을 충실히 구현함."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "요구된 시선 교환과 깁스 상태는 잘 반영되었으나, 군중 속 일부 인물의 형태와 질감이 다소 부자연스러움."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "레퍼런스에 존재하지 않는 천막, 테이블 및 제복 경찰관들을 배경에 임의로 추가하여 구조물/인물 창조 하드 위반."
   }
  ],
  "refs": [
   {
    "label": "SHOT BACKGROUND",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S47sh10__bgfirst_bg.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "왼쪽 무릎을 꿇은 형사의 자세가 레퍼런스 마네킹과 완전히 다릅니다(오른쪽 무릎을 꿇고 있음).",
     "fix_en": "Change the kneeling man's pose to match the reference exactly: place his left knee on the ground, raise his right knee, and place his left hand touching the ground.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "프롬프트에 명시된 훼손된 시체가 폴리스 라인 안쪽 바닥에 없습니다.",
     "fix_en": "Add the mutilated corpse lying on the ground inside the police tape, near the kneeling detective.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the kneeling man's pose to match the reference exactly: place his left knee on the ground, raise his right knee, and place his left hand touching the ground.\n- Add the mutilated corpse lying on the ground inside the police tape, near the kneeling detective.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S47sh10__bgfirst_bg.png",
   "bg_asset_id": "b667c132-a700-48ba-aaeb-f5f3496a2817",
   "bg_record_key": "S47sh10::bgfirst_bg",
   "chain_winner": true,
   "authority": "lane_conti_only"
  },
  "ref_mode": "lane(map_marker): 스케치+prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S47sh7"
  }
 },
 "S48sh3::bgfirst_bg": {
  "input_fingerprint": "48cf1c849cfa1260",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 식당 구석 테이블에 잔뜩 웅크리고 앉은 김지후(강수리영)가 주방 쪽의 고산들을 불안한 눈빛으로 곁눈질하는 측면.\n\nLOCATION (lock): At a corner table inside the island restaurant, facing its active kitchen in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At seated shoulder height beside the corner table, the lateral move settles into a medium profile of 김지후(강수리영), compressed into the left foreground with her shoulders drawn inward. The kitchen lies along her sidelong eyeline in the right background, where 고산들 remains occupied with the dead goat rather than acknowledging her stare.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, foreground, looks toward 고산들 in the kitchen; 고산들 in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 구석 테이블 (김지후(강수리영)가 앉아 있음) — One edge runs across the lower foreground in front of 김지후(강수리영); used as The table edge encloses 김지후(강수리영)'s defensive posture in the foreground; 주방 (고산들이 염소 고기를 손질하는 중) — The working area faces the camera obliquely beyond 김지후(강수리영)'s profile; used as The kitchen provides the distant endpoint of her sidelong gaze; 죽은 검은 염소와 핏빛 육고기 (머리가 잘렸으며 고산들이 손질하는 중); used as The dead animal and its butchered meat provide disturbing evidence behind 고산들 without overwhelming the frame.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 식당 구석 테이블에 잔뜩 웅크리고 앉은 김지후(강수리영)가 주방 쪽의 고산들을 불안한 눈빛으로 곁눈질하는 측면.\n\nLOCATION (lock): At a corner table inside the island restaurant, facing its active kitchen in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At seated shoulder height beside the corner table, the lateral move settles into a medium profile of 김지후(강수리영), compressed into the left foreground with her shoulders drawn inward. The kitchen lies along her sidelong eyeline in the right background, where 고산들 remains occupied with the dead goat rather than acknowledging her stare.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, foreground, looks toward 고산들 in the kitchen; 고산들 in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 구석 테이블 (김지후(강수리영)가 앉아 있음) — One edge runs across the lower foreground in front of 김지후(강수리영); used as The table edge encloses 김지후(강수리영)'s defensive posture in the foreground; 주방 (고산들이 염소 고기를 손질하는 중) — The working area faces the camera obliquely beyond 김지후(강수리영)'s profile; used as The kitchen provides the distant endpoint of her sidelong gaze; 죽은 검은 염소와 핏빛 육고기 (머리가 잘렸으며 고산들이 손질하는 중); used as The dead animal and its butchered meat provide disturbing evidence behind 고산들 without overwhelming the frame.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S48sh3__bgfirst_bg.png",
  "asset_id": "b6e29575-2c01-47a8-a329-e4bfe156a959",
  "input_asset_ids": [
   "900dee4b-18c0-44b7-bb0f-ef23a7fc6c12",
   "81ef185c-4c25-41b8-ad18-10745e6f506f"
  ]
 },
 "S48sh3": {
  "input_fingerprint": "13a4be7aab18bcf5",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 식당 구석 테이블에 잔뜩 웅크리고 앉은 김지후(강수리영)가 주방 쪽의 고산들을 불안한 눈빛으로 곁눈질하는 측면.\n\nLOCATION (lock): At a corner table inside the island restaurant, facing its active kitchen in daylight. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The black goat brought from the forest is now dead and decapitated, its bloody carcass being butchered. Jihu’s pendant remains missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 식당 구석 테이블에 잔뜩 웅크리고 앉은 김지후(강수리영)가 주방 쪽의 고산들을 불안한 눈빛으로 곁눈질하는 측면.\n\nLOCATION (lock): At a corner table inside the island restaurant, facing its active kitchen in daylight. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The black goat brought from the forest is now dead and decapitated, its bloody carcass being butchered. Jihu’s pendant remains missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 식당 구석 테이블에 잔뜩 웅크리고 앉은 김지후(강수리영)가 주방 쪽의 고산들을 불안한 눈빛으로 곁눈질하는 측면.\n\nLOCATION (lock): At a corner table inside the island restaurant, facing its active kitchen in daylight. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The black goat brought from the forest is now dead and decapitated, its bloody carcass being butchered. Jihu’s pendant remains missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S48sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S48sh3.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_mainland_harbor_dock_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "잔뜩 웅크린 자세로 주방 쪽을 돌아보며 불안해하는 측면 프레이밍을 잘 구현했으나, 주방의 인물이 단수로 묘사된 점이 아쉽습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "도축 중인 다수의 인물과 장소의 분위기는 좋으나, 주인공이 주방 쪽을 곁눈질하는 핵심적인 시선 행동이 전혀 일치하지 않습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "잔뜩 웅크린 자세로 주방 쪽을 돌아보며 불안해하는 측면 프레이밍을 잘 구현했으나, 주방의 인물이 단수로 묘사된 점이 아쉽습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "도축 중인 다수의 인물과 장소의 분위기는 좋으나, 주인공이 주방 쪽을 곁눈질하는 핵심적인 시선 행동이 전혀 일치하지 않습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "인물이 주방을 등지고 앉아 시선 방향(주방 쪽을 곁눈질)을 어겼으나, 시각적·물리적 오류가 없어 우선순위 규정에 따라 승리함."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "주방에서 도축 중인 남자의 왼팔이 아래로 향해 있음에도 고기 위에 뜬금없는 세 번째 손이 그려져 심각한 인체 구조 오류로 실격."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "인물이 주방을 등지고 앉아 시선 방향(주방 쪽을 곁눈질)을 어겼으나, 시각적·물리적 오류가 없어 우선순위 규정에 따라 승리함."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "주방에서 도축 중인 남자의 왼팔이 아래로 향해 있음에도 고기 위에 뜬금없는 세 번째 손이 그려져 심각한 인체 구조 오류로 실격."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 9,
     "B": 10
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 9,
   "B": 10
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "잔뜩 웅크린 자세로 주방 쪽을 돌아보며 불안해하는 측면 프레이밍을 잘 구현했으나, 주방의 인물이 단수로 묘사된 점이 아쉽습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "도축 중인 다수의 인물과 장소의 분위기는 좋으나, 주인공이 주방 쪽을 곁눈질하는 핵심적인 시선 행동이 전혀 일치하지 않습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_mainland_harbor_dock_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "김지후의 시선이 주방을 곁눈질하지 않고 반대 방향(오른쪽)을 향하고 있습니다.",
     "fix_en": "Shift Jihu's pupils to look sharply to the left, making her glance sideways at the kitchen.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "지후가 '잔뜩 웅크리고 앉은' 상태가 아니라 비교적 평범한 자세로 앉아 있습니다.",
     "fix_en": "Heavily hunch Jihu's shoulders forward and lower her head closer to the table so she appears tightly huddled.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Shift Jihu's pupils to look sharply to the left, making her glance sideways at the kitchen.\n- Heavily hunch Jihu's shoulders forward and lower her head closer to the table so she appears tightly huddled.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S48sh3__bgfirst_bg.png",
   "bg_asset_id": "b6e29575-2c01-47a8-a329-e4bfe156a959",
   "bg_record_key": "S48sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S48sh10": {
  "input_fingerprint": "c3da9adab6ba0be0",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)이 다정한 미소를 띤 채 김지후(강수리영)의 굳은 어깨를 손으로 감싸 쥐고 있는 모습.\n\nLOCATION (lock): At the corner dining table inside the restaurant, near the kitchen and adjoining bedroom. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the restaurant interior, kitchen sightline, table, daylight, and the younger woman's seated position. Exclude the earlier distant elder and show her hand resting on the woman's shoulder.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The slaughtered goat’s carcass and freshly cut meat remain in the kitchen; Jihu’s pendant remains missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 6,
   "A": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "젊은 여성의 고정된 시선과 자세, 식탁 위 사물(식판, 숟가락, 물컵) 등 레퍼런스의 요소를 완벽히 유지하면서 노년 여성의 다정한 동작을 매우 자연스럽게 추가했습니다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "요구사항을 전반적으로 잘 따랐으나, 식판의 구조와 물컵의 형태가 원본과 미세하게 다르며 어깨를 감싼 손의 위치가 다소 어색합니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "원본에서 고정되어야 할 젊은 여성의 시선과 고개 방향이 위로 변경되었으며, 식탁 위의 숟가락이 누락되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S48sh3_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "백발 여자가 젊은 여자의 어깨에 얹은 오른손의 앞쪽으로 5개의 손가락이 모두 감싸고 있어 손가락이 총 6개인 해부학적 오류가 있습니다.",
     "fix_en": "Remove the lowest extra finger from the hand resting on the shoulder, leaving only four fingers wrapping over the front.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the lowest extra finger from the hand resting on the shoulder, leaving only four fingers wrapping over the front.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S48sh3"
  }
 },
 "S49sh6": {
  "input_fingerprint": "81d9b4d64771d9e6",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 정인우가 내민 핸드폰 화면에 어둑한 골목길의 강영란(흰 머리 여자)과 황봉업(황 씨)이 찍힌 CCTV 캡쳐 사진이 선명하게 떠 있는 찰나.\n\nLOCATION (lock): At an outdoor table of a mainland seaside restaurant, overlooking the dark sea and distant fishing-boat lights. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast while he displays the CCTV capture on his phone.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 3,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "레퍼런스 구도를 복사하지 말라는 지시를 준수하며 야외 테이블 공간을 영리하게 재구성했고, 요구된 깁스와 CCTV 화면을 충실히 묘사함."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "CCTV 화면과 깁스는 표현되었으나, 레퍼런스 사진의 카메라 구도를 그대로 복사하여 '구도 복사 금지' 지시를 정면으로 위반함."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "인물이 폰을 내미는 구도는 훌륭하나, 레퍼런스의 건축적 특징을 누락했고 CCTV 화면에 금지된 텍스트(타임스탬프)가 노출됨."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L18B05.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "핸드폰 전면 화면 왼쪽 가장자리에 기기 후면에 있어야 할 카메라 렌즈 모듈이 잘못 생성되어 있습니다.",
     "fix_en": "Remove the rear camera lens module from the left side of the phone's front screen, replacing it with a normal flat display or black bezel.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the rear camera lens module from the left side of the phone's front screen, replacing it with a normal flat display or black bezel.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S49sh8": {
  "input_fingerprint": "4e7181c12b206889",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 조 형사와 신 형사가 놀란 표정으로 서로의 눈을 맞춘 찰나.\n\nLOCATION (lock): At the same outdoor seaside dining table under nighttime restaurant lighting. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the night-sea dining area, table, distant fishing-boat lights, and drinking setup. Exclude the phone-screen emphasis and show the two detectives exchanging a startled look.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo still wears the cast on his injured arm.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 3,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지정된 인물 레퍼런스와 깁스 설정을 정확히 반영하며 두 인물의 놀란 표정을 성공적으로 연출함."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "인물 신체가 허공에 떠 있는 심각한 콜라주 및 구조적 오류가 발생하여 핵심 규정을 위반함."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "지정된 인물(강영란)의 외모가 전혀 반영되지 않았으며 우측 인물의 양팔에 모두 깁스가 생성되는 오류가 있음."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S49sh6_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1445120>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 컷에서 폰을 쥐고 있던 남자의 오른팔에 있던 깁스가 이 사진에서는 왼팔로 위치가 바뀌어 연속성에 어긋납니다.",
     "fix_en": "Move the white cast from the man's left arm to his right arm.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "샷 텍스트는 두 명의 형사(조 형사와 신 형사)를 지시하고 있으나, 텍스트에 없는 인물(강영란 레퍼런스)이 등장했습니다.",
     "fix_en": "Replace the older woman on the left with a detective to align with the shot text.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "샷 텍스트에서 '놀란 표정'을 요구했으나, 왼쪽 여성의 표정은 차분하고 놀란 기색이 없습니다.",
     "fix_en": "Modify the woman's facial expression to show a startled look with widened eyes.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "텍스트 생성 금지 지침에도 불구하고 테이블 위 소주병 라벨에 글자가 선명하게 생성되었습니다.",
     "fix_en": "Remove the text from the label of the green bottle, leaving it blank.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Move the white cast from the man's left arm to his right arm.\n- Replace the older woman on the left with a detective to align with the shot text.\n- Modify the woman's facial expression to show a startled look with widened eyes.\n- Remove the text from the label of the green bottle, leaving it blank.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S49sh6"
  }
 },
 "S50sh2::bgfirst_bg": {
  "input_fingerprint": "396d785079a5dc9d",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 커튼 틈 사이로 보이는 골목길에 금월도 마을 청년들이 식당 쪽을 뚫어지게 주시하며 서 있는 전경.\n\nLOCATION (lock): Inside the dark rear bedroom of the island restaurant, at a curtained window overlooking the village lane.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level immediately beside the window, a wider lens peers through the narrow curtain opening, retaining both curtain edges as soft near-frame boundaries. The village youths appear full-bodied across the alley in the middle distance, spaced so their fixed attention toward the restaurant reads clearly without exposing the watcher inside.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: narrow curtain opening in the middle-center of the frame, foreground; village youths in the middle-center of the frame, midground, looks toward restaurant window.\n- KEY BACKGROUND ELEMENTS: 커튼 (Slightly parted) — The inner edges face the lens and leave only a narrow vertical opening; used as Forms dark near-frame boundaries on both sides and restricts the surveillance view; 창문 (Viewed through from inside) — The camera is aligned through the window toward the youths outside; no surface condition is specified; used as The lens looks through it from the back room toward the alley; 골목길 (Occupied by watching youths) — The visible stretch runs laterally beyond the window opening; used as Holds the watched figures and establishes their distance from the restaurant.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 커튼 틈 사이로 보이는 골목길에 금월도 마을 청년들이 식당 쪽을 뚫어지게 주시하며 서 있는 전경.\n\nLOCATION (lock): Inside the dark rear bedroom of the island restaurant, at a curtained window overlooking the village lane.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level immediately beside the window, a wider lens peers through the narrow curtain opening, retaining both curtain edges as soft near-frame boundaries. The village youths appear full-bodied across the alley in the middle distance, spaced so their fixed attention toward the restaurant reads clearly without exposing the watcher inside.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: narrow curtain opening in the middle-center of the frame, foreground; village youths in the middle-center of the frame, midground, looks toward restaurant window.\n- KEY BACKGROUND ELEMENTS: 커튼 (Slightly parted) — The inner edges face the lens and leave only a narrow vertical opening; used as Forms dark near-frame boundaries on both sides and restricts the surveillance view; 창문 (Viewed through from inside) — The camera is aligned through the window toward the youths outside; no surface condition is specified; used as The lens looks through it from the back room toward the alley; 골목길 (Occupied by watching youths) — The visible stretch runs laterally beyond the window opening; used as Holds the watched figures and establishes their distance from the restaurant.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S50sh2__bgfirst_bg.png",
  "asset_id": "ab196388-a35b-4d8b-b7b8-5d27bf1f2069",
  "input_asset_ids": [
   "79f5858a-0c1c-42bd-b8a3-65d344ad4515",
   "8cd56ef5-e3df-4d63-b2fb-2184a10e8936"
  ]
 },
 "S50sh2": {
  "input_fingerprint": "28c94085b36a2db5",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 커튼 틈 사이로 보이는 골목길에 금월도 마을 청년들이 식당 쪽을 뚫어지게 주시하며 서 있는 전경.\n\nLOCATION (lock): Inside the dark rear bedroom of the island restaurant, at a curtained window overlooking the village lane. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 커튼 틈 사이로 보이는 골목길에 금월도 마을 청년들이 식당 쪽을 뚫어지게 주시하며 서 있는 전경.\n\nLOCATION (lock): Inside the dark rear bedroom of the island restaurant, at a curtained window overlooking the village lane. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 커튼 틈 사이로 보이는 골목길에 금월도 마을 청년들이 식당 쪽을 뚫어지게 주시하며 서 있는 전경.\n\nLOCATION (lock): Inside the dark rear bedroom of the island restaurant, at a curtained window overlooking the village lane. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S50sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S50sh2.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_seaside_restaurant_terrace_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "커튼 틈 사이로 골목길에 서서 주시하는 청년들의 모습을 자연스럽게 연출했으며, 지시된 구도와 샷의 분위기를 잘 반영했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "창문 왼편에 레퍼런스 이미지의 외벽과 창문이 그대로 나타나는 물리적으로 불가능한 공간 구조 오류가 있으며 군중의 인체 구조도 부자연스럽습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "커튼 틈 사이로 골목길에 서서 주시하는 청년들의 모습을 자연스럽게 연출했으며, 지시된 구도와 샷의 분위기를 잘 반영했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "창문 왼편에 레퍼런스 이미지의 외벽과 창문이 그대로 나타나는 물리적으로 불가능한 공간 구조 오류가 있으며 군중의 인체 구조도 부자연스럽습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "시야 좌측 창문에 밖에서 봐야 할 식당 외벽과 창문이 그대로 합성되어 물리적으로 불가능한 뷰를 연출한 치명적인 하드 비올레이션입니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "커튼 틈 사이로 내다보는 1인칭 구도를 정확히 살렸으며, 레퍼런스 골목길의 디테일과 식당 쪽을 주시하는 청년들의 모습을 프롬프트대로 훌륭하게 구현했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "시야 좌측 창문에 밖에서 봐야 할 식당 외벽과 창문이 그대로 합성되어 물리적으로 불가능한 뷰를 연출한 치명적인 하드 비올레이션입니다."
     },
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "커튼 틈 사이로 내다보는 1인칭 구도를 정확히 살렸으며, 레퍼런스 골목길의 디테일과 식당 쪽을 주시하는 청년들의 모습을 프롬프트대로 훌륭하게 구현했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 6
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "커튼 틈 사이로 골목길에 서서 주시하는 청년들의 모습을 자연스럽게 연출했으며, 지시된 구도와 샷의 분위기를 잘 반영했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "창문 왼편에 레퍼런스 이미지의 외벽과 창문이 그대로 나타나는 물리적으로 불가능한 공간 구조 오류가 있으며 군중의 인체 구조도 부자연스럽습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_seaside_restaurant_terrace_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "레이아웃 스케치에는 총 5명의 인물이 그려져 있으나, 생성된 이미지에는 우측의 한 명이 누락되어 4명만 있습니다.",
     "fix_en": "Add the missing fifth person on the right side of the alley, exactly as positioned in the layout sketch.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "레이아웃 스케치에서 우측 인물들은 팔을 아래로 내리고 있으나, 생성된 이미지의 우측 두 남자는 팔짱을 끼고 있어 포즈가 일치하지 않습니다.",
     "fix_en": "Uncross the arms of the two men on the right, hanging their arms down by their sides to match the poses in the layout sketch.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the missing fifth person on the right side of the alley, exactly as positioned in the layout sketch.\n- Uncross the arms of the two men on the right, hanging their arms down by their sides to match the poses in the layout sketch.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S50sh2__bgfirst_bg.png",
   "bg_asset_id": "ab196388-a35b-4d8b-b7b8-5d27bf1f2069",
   "bg_record_key": "S50sh2::bgfirst_bg",
   "chain_winner": true,
   "authority": "seed_bg"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S50sh3": {
  "input_fingerprint": "743ef2ef95e97e03",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 커튼을 꽉 쥔 채 긴장된 표정으로 방 안을 향해 돌아선 찰나.\n\nLOCATION (lock): Within the unlit rear bedroom, immediately inside the curtained window. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the dark bedroom, curtain gap, window, and youths visible in the alley outside. Exclude the outward-looking pose and show the woman turning back into the room.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 5,
   "A": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "이전 샷의 인물들(창밖 청년들)과 창틀의 형태를 완벽하게 유지하면서, 커튼을 잡고 방 안을 향해 긴장된 표정으로 돌아보는 모습을 가장 충실하게 연출했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "커튼을 꽉 쥔 행동과 표정은 훌륭하나, 창밖 청년들 중 일부가 누락되거나 복장이 변경되어 '이전 샷의 상태 유지' 지침을 어겼습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "이전 샷에서 반드시 포함해야 할 창밖 골목의 청년들이 완전히 누락되었고, 방 안을 향해 돌아보는 시선 방향도 맞지 않는 심각한 오류가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S50sh2_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "창틀 중앙을 잡고 있는 여성의 오른손이 비정상적으로 작고 팔과 분리된 것처럼 보여 해부학적으로 왜곡되어 있습니다.",
     "fix_en": "Redraw the woman's right hand gripping the central window frame to have correct human anatomy and proper proportions, ensuring it connects naturally to her sleeve.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the woman's right hand gripping the central window frame to have correct human anatomy and proper proportions, ensuring it connects naturally to her sleeve.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S50sh2"
  }
 },
 "S51sh2::bgfirst_bg": {
  "input_fingerprint": "4ecabd1e54814009",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 고산들이 바닥에 묶인 강민숙(홍은미)을 안타까운 시선으로 묵묵히 내려다보는 전신.\n\nLOCATION (lock): Deep inside a cave prison, weakly illuminated by a low-wattage electric bulb.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From standing chest height on the prescribed oblique axis, a static wide frame includes 고산들 at full height in one side of the midground and the chained 강민숙 low on the cave floor opposite him. Their vertical separation carries the emotional imbalance, while enough open space remains between them for the later forward move along their shared eyeline.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 고산들 in the middle-left of the frame, midground, looks toward bound Minsuk; 강민숙 in the lower-right of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 쇠사슬 (Binding 강민숙) — The chain runs from 강민숙’s restrained body toward its off-frame securing point; used as Physically confines 강민숙 and makes her restrained condition legible in the wide frame; 촉 낮은 전구 등 (On) — The lit bulb fixture is positioned within the cave space above the subjects; used as Visible illumination source established by the scene; 동굴 내부 (Occupied by the two characters); used as Defines the enclosed setting around the separated figures.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 고산들이 바닥에 묶인 강민숙(홍은미)을 안타까운 시선으로 묵묵히 내려다보는 전신.\n\nLOCATION (lock): Deep inside a cave prison, weakly illuminated by a low-wattage electric bulb.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From standing chest height on the prescribed oblique axis, a static wide frame includes 고산들 at full height in one side of the midground and the chained 강민숙 low on the cave floor opposite him. Their vertical separation carries the emotional imbalance, while enough open space remains between them for the later forward move along their shared eyeline.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 고산들 in the middle-left of the frame, midground, looks toward bound Minsuk; 강민숙 in the lower-right of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 쇠사슬 (Binding 강민숙) — The chain runs from 강민숙’s restrained body toward its off-frame securing point; used as Physically confines 강민숙 and makes her restrained condition legible in the wide frame; 촉 낮은 전구 등 (On) — The lit bulb fixture is positioned within the cave space above the subjects; used as Visible illumination source established by the scene; 동굴 내부 (Occupied by the two characters); used as Defines the enclosed setting around the separated figures.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S51sh2__bgfirst_bg.png",
  "asset_id": "d9b22cf7-fbc9-456e-8a40-b9ee9dc6bb5f",
  "input_asset_ids": [
   "cee37974-5667-44a5-b263-c37121c22b76",
   "9379f746-9fec-4d7c-bca5-94d3e636b113"
  ]
 },
 "S51sh2": {
  "input_fingerprint": "e1481eac0eff1f01",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 고산들이 바닥에 묶인 강민숙(홍은미)을 안타까운 시선으로 묵묵히 내려다보는 전신.\n\nLOCATION (lock): Deep inside a cave prison, weakly illuminated by a low-wattage electric bulb. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is seated low on the cave floor with her weakened torso slumped and her limbs restrained by heavy chains, positioned directly beneath Sandul as he looks down at her.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk remains shackled in the cave, unable to leave.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 고산들이 바닥에 묶인 강민숙(홍은미)을 안타까운 시선으로 묵묵히 내려다보는 전신.\n\nLOCATION (lock): Deep inside a cave prison, weakly illuminated by a low-wattage electric bulb. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is seated low on the cave floor with her weakened torso slumped and her limbs restrained by heavy chains, positioned directly beneath Sandul as he looks down at her.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk remains shackled in the cave, unable to leave.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 고산들이 바닥에 묶인 강민숙(홍은미)을 안타까운 시선으로 묵묵히 내려다보는 전신.\n\nLOCATION (lock): Deep inside a cave prison, weakly illuminated by a low-wattage electric bulb. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is seated low on the cave floor with her weakened torso slumped and her limbs restrained by heavy chains, positioned directly beneath Sandul as he looks down at her.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk remains shackled in the cave, unable to leave.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S51sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S51sh2.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L30B01.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "요청된 전신 샷(풀샷)을 잘 담아냈으며, 남자가 여자 바로 곁에 서서 내려다보는 지정된 위치 관계를 충실히 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "전신 샷은 구현되었으나, 두 인물 간의 거리가 너무 멀어 여자가 남자 '바로 아래'에 있어야 한다는 지시를 위반했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "요청된 전신 샷(풀샷)을 잘 담아냈으며, 남자가 여자 바로 곁에 서서 내려다보는 지정된 위치 관계를 충실히 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "전신 샷은 구현되었으나, 두 인물 간의 거리가 너무 멀어 여자가 남자 '바로 아래'에 있어야 한다는 지시를 위반했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "전신 샷의 프레이밍과 인물들의 자세가 지시문에 잘 부합하며, 사슬과 신체가 중력을 거스르지 않고 자연스럽게 묘사되었습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "문손잡이에 연결된 쇠사슬이 중력을 무시한 채 공중에 일직선으로 떠 있는 물리적으로 불가능한 묘사(Hard Violation)가 발생하여 탈락입니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "전신 샷의 프레이밍과 인물들의 자세가 지시문에 잘 부합하며, 사슬과 신체가 중력을 거스르지 않고 자연스럽게 묘사되었습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "문손잡이에 연결된 쇠사슬이 중력을 무시한 채 공중에 일직선으로 떠 있는 물리적으로 불가능한 묘사(Hard Violation)가 발생하여 탈락입니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 7,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 7,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "요청된 전신 샷(풀샷)을 잘 담아냈으며, 남자가 여자 바로 곁에 서서 내려다보는 지정된 위치 관계를 충실히 구현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "전신 샷은 구현되었으나, 두 인물 간의 거리가 너무 멀어 여자가 남자 '바로 아래'에 있어야 한다는 지시를 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L30B01.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "강민숙의 상체는 오른쪽 벽에 기대어 있는데, 하반신(다리와 신발)은 방 한가운데에 비정상적으로 길게 늘어지거나 절단된 형태로 분리되어 있습니다.",
     "fix_en": "Remove the disconnected legs and boots lying in the center of the floor, and render Minsuk's legs properly attached to her torso, resting naturally on the floor by the right wall.",
     "unfixable": false
    },
    {
     "issue_ko": "강민숙이 산들의 시선 바로 아래(방 중앙)에 위치해야 한다는 지시와 달리, 오른쪽 문가의 벽 쪽에 멀리 떨어져 배치되어 있습니다.",
     "fix_en": "Kang Minsuk must be seated on the floor directly beneath Sandul in the center of the room.",
     "unfixable": true
    },
    {
     "issue_ko": "금지된 텍스트가 왼쪽 선반 아래쪽의 상자들 표면에 생성되어 있습니다.",
     "fix_en": "Remove all text and letter-like markings from the boxes on the left shelves, leaving the packaging surfaces blank.",
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the disconnected legs and boots lying in the center of the floor, and render Minsuk's legs properly attached to her torso, resting naturally on the floor by the right wall.\n- Remove all text and letter-like markings from the boxes on the left shelves, leaving the packaging surfaces blank.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S51sh2__bgfirst_bg.png",
   "bg_asset_id": "d9b22cf7-fbc9-456e-8a40-b9ee9dc6bb5f",
   "bg_record_key": "S51sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S51sh7": {
  "input_fingerprint": "962588ca636399e1",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 돌아선 고산들의 뒷모습을 향해 강민숙(홍은미)이 절박하게 애원하며 부르짖는 얼굴.\n\nLOCATION (lock): At the chained holding area deep inside the dim cave. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the dim cave, low bulb, rough stone, chains, and captive woman's position. Exclude the man's earlier downward gaze and show him turned away while she pleads.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is seated low on the cave floor with her weakened torso slumped and her limbs restrained by heavy chains, positioned directly beneath Sandul as he looks down at her.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk remains seated low on the cave floor with her weakened torso slumped and her limbs restrained by heavy chains, positioned directly beneath Sandeul as he looks down at her.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 6,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "남자가 돌아선 뒷모습과 여자가 애원하는 얼굴 표정을 가장 사실적이고 정확하게 구현했으며, 기준 이미지의 제약 조건들도 완벽하게 유지했습니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "지시된 상황과 인물의 배치를 잘 따랐으나, 여자의 절박하게 부르짖는 표정 묘사가 다소 흐릿하고 덜 자연스럽습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "프롬프트에 없는 남자가 한 명 더 추가되어 총 3명의 인물이 등장하는 치명적인 구조적 위반(Hard Violation)이 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S51sh2_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "샷 텍스트는 여자가 애원하는 '얼굴'을 지시하여 클로즈업을 요구하지만, 결과물은 이전 샷과 동일한 롱샷 구도를 그대로 유지하고 있습니다.",
     "fix_en": "Crop the image tightly around the woman's face and the man's back to simulate a closer shot scale.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "남자의 상체와 고개는 뒤돌아서 뒷벽을 향하고 있으나, 두 발과 신발은 여전히 오른쪽(여자 쪽)을 향하고 있어 신체가 해부학적으로 불가능하게 뒤틀려 있습니다.",
     "fix_en": "Redraw the man's boots and lower legs so the heels face the camera and the toes point toward the back wall.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "남자의 우측에 늘어뜨린 오른손의 형태가 뭉개져 단순한 덩어리처럼 보이며 손가락 마디가 묘사되지 않았습니다.",
     "fix_en": "Redraw the man's visible right hand to show anatomically correct, clearly separated fingers and knuckles.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the man's boots and lower legs so the heels face the camera and the toes point toward the back wall.\n- Redraw the man's visible right hand to show anatomically correct, clearly separated fingers and knuckles.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S51sh2"
  }
 },
 "S52sh2::bgfirst_bg": {
  "input_fingerprint": "c14169d0138bd82b",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 피 묻은 거대 바위 앞에 고산들이 무릎을 꿇은 채 앉아 있는 뒷모습.\n\nLOCATION (lock): In a forest clearing before a massive blood-smeared ritual boulder at night.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From adult eye height far behind 고산들, the static wide frame is centered precisely on the shared axis between his kneeling body and the enormous boulder. His back occupies the lower center as a small human scale reference, while the blood-marked rock rises through the central background without being cropped.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: kneeling 고산들 in the lower-center of the frame, foreground; blood-marked giant boulder in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 거대한 바위 (Marked in places with dark-red animal blood handprints) — Its broad blood-marked face is directed toward the camera behind 고산들; used as Central ritual anchor and dominant background mass, scaled against the kneeling figure; 숲속 공터 (Open clearing within the forest); used as Frames the boulder and kneeling figure within the established outdoor ritual space; 숲 (Night setting); used as Surrounds the clearing at the frame edges without interrupting the central axis.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 피 묻은 거대 바위 앞에 고산들이 무릎을 꿇은 채 앉아 있는 뒷모습.\n\nLOCATION (lock): In a forest clearing before a massive blood-smeared ritual boulder at night.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From adult eye height far behind 고산들, the static wide frame is centered precisely on the shared axis between his kneeling body and the enormous boulder. His back occupies the lower center as a small human scale reference, while the blood-marked rock rises through the central background without being cropped.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: kneeling 고산들 in the lower-center of the frame, foreground; blood-marked giant boulder in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 거대한 바위 (Marked in places with dark-red animal blood handprints) — Its broad blood-marked face is directed toward the camera behind 고산들; used as Central ritual anchor and dominant background mass, scaled against the kneeling figure; 숲속 공터 (Open clearing within the forest); used as Frames the boulder and kneeling figure within the established outdoor ritual space; 숲 (Night setting); used as Surrounds the clearing at the frame edges without interrupting the central axis.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S52sh2__bgfirst_bg.png",
  "asset_id": "3d44b98e-293d-4e3b-8cd3-c1ba135ed913",
  "input_asset_ids": [
   "fa06a65c-7e9a-4b7f-88b7-c35ca6c04609",
   "df8be972-4b25-458c-80ca-f856af22f0d3"
  ]
 },
 "S52sh2": {
  "input_fingerprint": "657d013eba8449ae",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 피 묻은 거대 바위 앞에 고산들이 무릎을 꿇은 채 앉아 있는 뒷모습.\n\nLOCATION (lock): In a forest clearing before a massive blood-smeared ritual boulder at night. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Dark-red animal-blood handprints and smears remain across the giant ritual rock.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 피 묻은 거대 바위 앞에 고산들이 무릎을 꿇은 채 앉아 있는 뒷모습.\n\nLOCATION (lock): In a forest clearing before a massive blood-smeared ritual boulder at night. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Dark-red animal-blood handprints and smears remain across the giant ritual rock.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 피 묻은 거대 바위 앞에 고산들이 무릎을 꿇은 채 앉아 있는 뒷모습.\n\nLOCATION (lock): In a forest clearing before a massive blood-smeared ritual boulder at night. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Dark-red animal-blood handprints and smears remain across the giant ritual rock.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S52sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S52sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:876576>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B04.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:876576>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "거대 바위의 스케일과 피사체 간의 공간감을 잘 살렸으며, 지시된 뒷모습과 무릎 꿇은 자세를 자연스럽게 연출했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "인물이 지나치게 크게 배치되어 '거대 바위'라는 지시사항의 스케일이 축소되어 보이며 공간감이 다소 부족합니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "거대 바위의 스케일과 피사체 간의 공간감을 잘 살렸으며, 지시된 뒷모습과 무릎 꿇은 자세를 자연스럽게 연출했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "인물이 지나치게 크게 배치되어 '거대 바위'라는 지시사항의 스케일이 축소되어 보이며 공간감이 다소 부족합니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 숏 크기와 뒷모습 앵글을 정확히 구현했으며, 레퍼런스의 재킷 디테일과 체격을 충실히 반영함."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "뒷모습과 무릎 꿇은 자세는 맞으나, 의상 디테일이 레퍼런스와 다르고 인물의 신체 비율과 발목 묘사가 다소 어색함."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 숏 크기와 뒷모습 앵글을 정확히 구현했으며, 레퍼런스의 재킷 디테일과 체격을 충실히 반영함."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "뒷모습과 무릎 꿇은 자세는 맞으나, 의상 디테일이 레퍼런스와 다르고 인물의 신체 비율과 발목 묘사가 다소 어색함."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 12
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 11,
   "B": 12
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "거대 바위의 스케일과 피사체 간의 공간감을 잘 살렸으며, 지시된 뒷모습과 무릎 꿇은 자세를 자연스럽게 연출했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "인물이 지나치게 크게 배치되어 '거대 바위'라는 지시사항의 스케일이 축소되어 보이며 공간감이 다소 부족합니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B04.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "로케이션 레퍼런스 이미지의 카메라 구도를 그대로 복사했습니다 ('Never copy its camera framing' 지시 위반).",
     "fix_en": "Change the camera angle, focal length, or framing so it does not exactly duplicate the location reference image.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "캐릭터가 무릎을 꿇은 자세에서 종아리 부분이 생략된 채 신발 바닥이 허벅지와 무릎 뒤에 바로 붙어 있어 인체 비례가 어색합니다.",
     "fix_en": "Redraw the lower legs and boots to show the correct anatomical length and perspective for a kneeling posture.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the lower legs and boots to show the correct anatomical length and perspective for a kneeling posture.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S52sh2__bgfirst_bg.png",
   "bg_asset_id": "3d44b98e-293d-4e3b-8cd3-c1ba135ed913",
   "bg_record_key": "S52sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S52sh4": {
  "input_fingerprint": "3442b9855b041808",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 고산들이 고개를 푹 숙인 채 간절한 표정으로 기도를 올리는 측면.\n\nLOCATION (lock): At the base of the blood-marked boulder in the open ritual clearing. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the forest clearing, blood-smeared boulder, dark night, and kneeling position. Exclude the rear-view emphasis and show the man's bowed praying profile.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The ritual rock remains marked with dark-red animal blood.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 5,
   "A": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "제시된 측면 앵글과 푹 고개 숙인 기도 자세를 완벽히 구현했으며, 인물과 배경의 연속성도 가장 우수합니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "측면 구도와 연속성은 양호하나, 맞잡은 손의 손가락들이 기형적으로 뭉개지는 치명적인 해부학적 오류가 있습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "레퍼런스에 없는 소품(돌탑, 띠)이 임의로 생성된 하드 위반이며, 인물이 바위를 등지고 있어 연속성도 깨졌습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S52sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지문은 '고개를 푹 숙인 채'라고 명시했으나, 인물의 고개가 살짝만 기울어져 있고 푹 숙인 상태가 아닙니다.",
     "fix_en": "Tilt the man's head much further down so his chin is tucked close to his chest in a deeply bowed position.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Tilt the man's head much further down so his chin is tucked close to his chest in a deeply bowed position.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S52sh2"
  }
 },
 "groupbg::육지_해안마을_선착장": {
  "input_fingerprint": "f69ceed56648a04e",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "육지_해안마을_선착장",
    "tags": [
     "S100sh3",
     "S100sh5",
     "S47sh1",
     "S47sh10",
     "S47sh7",
     "S53sh3",
     "S53sh5"
    ]
   },
   "context_sig": "133c9af8b04c9f2c"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On a mainland pier beside a moored fishing boat in daylight.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 숲속 동굴 내부: 빛이 들지 않는 깊고 축축한 자연 동굴. 바닥에 인골과 토막 난 시체가 널려 있다. (특징: 울퉁불퉁하고 어두운 암석 벽과 천장; 천장에 빽빽이 매달린 수백 마리 박쥐 떼; 바닥에 흩어진 사람의 뼈와 토막 난 신체 부위; 바닥에 고인 피 웅덩이; 벽에 고정된 무거운 쇠사슬) / 금월도 숲속 만월제 도당: 숲 한가운데 둥글게 트인 공터. 거대한 바위와 제단이 있고 커다란 모닥불들이 타오른다. (특징: 성인 키 서너 배 높이의 거대한 암석 바위; 피 손자국으로 범벅이 된 바위 표면; 바위 앞 제단; 제단을 둥글게 감싸며 타오르는 커다란 모닥불들; 하늘 위 거대한 핏빛 보름달) / 금월도 해송 숲길 및 숲속: 해송이 울창하게 우거져 빛이 잘 들지 않는 빽빽하고 좁은 흙길과 숲의 심층부. (특징: 빽빽한 해송 나무 기둥들; 무성하고 어두운 수풀; 바닥에 깔린 흙과 솔잎; ‘수리영’이 적힌 이끼 낀 비석들; 비석 앞에 놓인 썩은 핏빛 육고기 덩어리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 해안마을 선착장 - 낮\n- 실외. 해안마을 선착장 - 낮\n- 실외. 해안마을 선착장 - 낮\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On a mainland pier beside a moored fishing boat in daylight.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 숲속 동굴 내부: 빛이 들지 않는 깊고 축축한 자연 동굴. 바닥에 인골과 토막 난 시체가 널려 있다. (특징: 울퉁불퉁하고 어두운 암석 벽과 천장; 천장에 빽빽이 매달린 수백 마리 박쥐 떼; 바닥에 흩어진 사람의 뼈와 토막 난 신체 부위; 바닥에 고인 피 웅덩이; 벽에 고정된 무거운 쇠사슬) / 금월도 숲속 만월제 도당: 숲 한가운데 둥글게 트인 공터. 거대한 바위와 제단이 있고 커다란 모닥불들이 타오른다. (특징: 성인 키 서너 배 높이의 거대한 암석 바위; 피 손자국으로 범벅이 된 바위 표면; 바위 앞 제단; 제단을 둥글게 감싸며 타오르는 커다란 모닥불들; 하늘 위 거대한 핏빛 보름달) / 금월도 해송 숲길 및 숲속: 해송이 울창하게 우거져 빛이 잘 들지 않는 빽빽하고 좁은 흙길과 숲의 심층부. (특징: 빽빽한 해송 나무 기둥들; 무성하고 어두운 수풀; 바닥에 깔린 흙과 솔잎; ‘수리영’이 적힌 이끼 낀 비석들; 비석 앞에 놓인 썩은 핏빛 육고기 덩어리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 해안마을 선착장 - 낮\n- 실외. 해안마을 선착장 - 낮\n- 실외. 해안마을 선착장 - 낮\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_육지_해안마을_선착장_984857.png",
  "asset_id": "db14a041-c002-4b4d-b616-3a9abbd23a9a",
  "input_asset_ids": [
   "155d5e8e-7eb3-46d5-bb47-6074ee060d35"
  ],
  "origin_tag": "S53sh3",
  "place_text": "On a mainland pier beside a moored fishing boat in daylight.",
  "origin_inputs": {
   "place_text": "On a mainland pier beside a moored fishing boat in daylight.",
   "time_of_day_en": "day",
   "conti_asset_id": "155d5e8e-7eb3-46d5-bb47-6074ee060d35"
  }
 },
 "S53sh3::bgfirst_bg": {
  "input_fingerprint": "2551de11f14354e9",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 인상을 잔뜩 구긴 황봉업(황 씨)이 정인우를 향해 한 손을 바깥쪽으로 뻗어 거절하는 자세로 멈춰 있는 찰나.\n\nLOCATION (lock): On a mainland pier beside a moored fishing boat in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Tracking from the dock-facing side of the boat, the camera finishes at 황봉업’s upper-body height in a medium two-shot, with him closer to frame center and 정인우 held across the confrontation axis near the opposite edge. 황봉업’s outward palm arrests the visual line between them while 정인우’s forward-set posture keeps the refusal from reading as casual conversation.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 어선 (At the dock) — The dock-facing side of the vessel is visible behind and beneath 황봉업; used as Places 황봉업 at a different level and establishes the physical boundary of the confrontation; 선착장 (Daytime setting) — The dock extends along the camera side of the boat; used as Defines the exterior confrontation space and camera side.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 인상을 잔뜩 구긴 황봉업(황 씨)이 정인우를 향해 한 손을 바깥쪽으로 뻗어 거절하는 자세로 멈춰 있는 찰나.\n\nLOCATION (lock): On a mainland pier beside a moored fishing boat in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Tracking from the dock-facing side of the boat, the camera finishes at 황봉업’s upper-body height in a medium two-shot, with him closer to frame center and 정인우 held across the confrontation axis near the opposite edge. 황봉업’s outward palm arrests the visual line between them while 정인우’s forward-set posture keeps the refusal from reading as casual conversation.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 어선 (At the dock) — The dock-facing side of the vessel is visible behind and beneath 황봉업; used as Places 황봉업 at a different level and establishes the physical boundary of the confrontation; 선착장 (Daytime setting) — The dock extends along the camera side of the boat; used as Defines the exterior confrontation space and camera side.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S53sh3__bgfirst_bg.png",
  "asset_id": "6734259d-0bb7-4d1f-bc8d-56dde48c3c5d",
  "input_asset_ids": [
   "155d5e8e-7eb3-46d5-bb47-6074ee060d35",
   "db14a041-c002-4b4d-b616-3a9abbd23a9a"
  ]
 },
 "S53sh3": {
  "input_fingerprint": "bcb5860af39799c6",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 인상을 잔뜩 구긴 황봉업(황 씨)이 정인우를 향해 한 손을 바깥쪽으로 뻗어 거절하는 자세로 멈춰 있는 찰나.\n\nLOCATION (lock): On a mainland pier beside a moored fishing boat in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 인상을 잔뜩 구긴 황봉업(황 씨)이 정인우를 향해 한 손을 바깥쪽으로 뻗어 거절하는 자세로 멈춰 있는 찰나.\n\nLOCATION (lock): On a mainland pier beside a moored fishing boat in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 인상을 잔뜩 구긴 황봉업(황 씨)이 정인우를 향해 한 손을 바깥쪽으로 뻗어 거절하는 자세로 멈춰 있는 찰나.\n\nLOCATION (lock): On a mainland pier beside a moored fishing boat in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S53sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S53sh3.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_육지_해안마을_선착장_984857.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 부둣가 배경에서 깁스 설정과 인상을 구기며 손을 뻗어 거절하는 자세(손바닥을 밖으로 향함)를 매우 정확하게 연출했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "배경과 깁스 설정은 일치하나, 손을 뻗은 자세가 명확한 '거절'의 제스처라기보다는 일반적인 손짓에 가까워 프롬프트 충실도가 다소 떨어집니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 부둣가 배경에서 깁스 설정과 인상을 구기며 손을 뻗어 거절하는 자세(손바닥을 밖으로 향함)를 매우 정확하게 연출했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "배경과 깁스 설정은 일치하나, 손을 뻗은 자세가 명확한 '거절'의 제스처라기보다는 일반적인 손짓에 가까워 프롬프트 충실도가 다소 떨어집니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "황봉업의 잔뜩 구긴 표정과 정인우의 깁스 등 주요 설정을 잘 반영했으며, 인물들의 해부학적 구조와 위치 관계가 자연스럽습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "거절하는 제스처와 표정의 연출은 훌륭하나, 황봉업의 뻗은 왼팔에 오른손이 달려 있고 손바닥 방향이 어색한 치명적인 해부학적 오류가 있습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "황봉업의 잔뜩 구긴 표정과 정인우의 깁스 등 주요 설정을 잘 반영했으며, 인물들의 해부학적 구조와 위치 관계가 자연스럽습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "거절하는 제스처와 표정의 연출은 훌륭하나, 황봉업의 뻗은 왼팔에 오른손이 달려 있고 손바닥 방향이 어색한 치명적인 해부학적 오류가 있습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 10,
     "B": 12
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 10,
   "B": 12
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지정된 부둣가 배경에서 깁스 설정과 인상을 구기며 손을 뻗어 거절하는 자세(손바닥을 밖으로 향함)를 매우 정확하게 연출했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "배경과 깁스 설정은 일치하나, 손을 뻗은 자세가 명확한 '거절'의 제스처라기보다는 일반적인 손짓에 가까워 프롬프트 충실도가 다소 떨어집니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_육지_해안마을_선착장_984857.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "황 씨가 뻗은 오른손의 엄지손가락이 아래를 향하고 있어 손의 해부학적 구조가 틀림.",
     "fix_en": "Redraw the extended right hand to be anatomically correct, with the thumb pointing up naturally.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "배의 난간에서 뻗어 나온 얇은 흰색 선이 정인우의 팔 깁스를 부자연스럽게 관통하고 있음.",
     "fix_en": "Remove the thin white line that incorrectly connects the boat's railing to the cast.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the extended right hand to be anatomically correct, with the thumb pointing up naturally.\n- Remove the thin white line that incorrectly connects the boat's railing to the cast.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S53sh3__bgfirst_bg.png",
   "bg_asset_id": "6734259d-0bb7-4d1f-bc8d-56dde48c3c5d",
   "bg_record_key": "S53sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "육지_해안마을_선착장",
   "groupbg_asset_id": "db14a041-c002-4b4d-b616-3a9abbd23a9a"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S53sh5": {
  "input_fingerprint": "1285815fa3eb283b",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 정인우가 팔을 앞으로 곧게 뻗어 경찰 신분증을 황봉업(황 씨)의 눈앞에 바짝 들이댄 채 정지해 있는 찰나의 클로즈업.\n\nLOCATION (lock): At the edge of the mainland pier, face to face with the operator aboard the moored fishing boat. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the daylight harbor, fishing boat, pier materials, and tense face-to-face positions. Exclude the boatman's refusal gesture and emphasize the police identification held before him.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo still wears his arm cast while presenting his police identification.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 3,
   "A": 2
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "신분증의 앞면이 상대방이 아닌 카메라를 향하는 소품 방향 오류가 있으나, 유일하게 해부학적 오류가 없고 '눈앞에 바짝 들이댄' 지문을 완벽하게 연출했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "오른팔 깁스에서 뻗어 나온 손이 왼손으로 묘사되는 심각한 해부학적 오류가 있으며, 신분증의 방향 또한 규칙에 어긋납니다."
   },
   {
    "label": "A",
    "score": 2,
    "verdict_ko": "깁스를 한 오른팔이 아래로 향해 있음에도 가슴 부위에서 신분증을 든 세 번째 팔/손이 등장하는 치명적인 신체 구조 위반이 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S53sh3_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "정인우의 오른팔이 슬링에 걸쳐진 상태로 아래에 있는데, 카드를 쥐고 있는 또 다른 깁스 한 오른팔이 나타나 팔이 3개가 되었습니다.",
     "fix_en": "Remove the arm and cast resting in the sling at the bottom left, keeping only the extended arm to correct the anatomy.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "경찰 신분증의 앞면이 황 씨의 눈이 아닌 카메라 쪽을 향하고 있어 기능적 방향이 잘못되었습니다.",
     "fix_en": "Rotate the ID card so its front surface faces directly toward the older man's eyes, exposing its back or edge to the camera.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "프롬프트에서 모든 텍스트 생성을 금지했으나, 신분증 표면에 한글 텍스트가 명확하게 적혀 있습니다.",
     "fix_en": "Erase all letters and readable text from the ID card.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the arm and cast resting in the sling at the bottom left, keeping only the extended arm to correct the anatomy.\n- Rotate the ID card so its front surface faces directly toward the older man's eyes, exposing its back or edge to the camera.\n- Erase all letters and readable text from the ID card.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S53sh3"
  }
 },
 "S54sh3::bgfirst_bg": {
  "input_fingerprint": "4a94d4882ba6b8b4",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 조타실 안에서 핸들을 쥔 황봉업(황 씨)이 뱃머리의 정인우를 향해 적대감 가득한 눈빛을 쏘아보는 측면.\n\nLOCATION (lock): Inside the wheelhouse of a fishing boat moving through dense daytime fog.\n\nTIME OF DAY (lock): day, dense fog.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From inside the wheelhouse at shoulder height, hold a static medium side view close to 황봉업(황 씨), with his clenched hands on the wheel in the near left and his hostile profile occupying the middle third. Keep 정인우 small but legible on the bow along 황봉업(황 씨)'s outward sightline, preserving the visual axis that will cut forward toward the island.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 황봉업(황 씨) in the middle-left of the frame, foreground, looks toward Jeong Inwoo on the bow; 정인우 in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 조타실 핸들 (Held firmly by 황봉업(황 씨)) — Its gripped rim is turned partly toward the camera while the central assembly recedes into the wheelhouse; used as Held beneath 황봉업(황 씨)'s hands as the immediate evidence of control over the boat; 어선 뱃머리 (Advancing through the sea) — Its forward-pointing side recedes away from the camera toward the sea; used as Extends through the background as the endpoint of 황봉업(황 씨)'s hostile sightline; 바다의 안개 (Dense enough to obscure distant visibility); used as Reduces background clarity around the bow and prepares the following reveal.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 조타실 안에서 핸들을 쥔 황봉업(황 씨)이 뱃머리의 정인우를 향해 적대감 가득한 눈빛을 쏘아보는 측면.\n\nLOCATION (lock): Inside the wheelhouse of a fishing boat moving through dense daytime fog.\n\nTIME OF DAY (lock): day, dense fog.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From inside the wheelhouse at shoulder height, hold a static medium side view close to 황봉업(황 씨), with his clenched hands on the wheel in the near left and his hostile profile occupying the middle third. Keep 정인우 small but legible on the bow along 황봉업(황 씨)'s outward sightline, preserving the visual axis that will cut forward toward the island.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 황봉업(황 씨) in the middle-left of the frame, foreground, looks toward Jeong Inwoo on the bow; 정인우 in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 조타실 핸들 (Held firmly by 황봉업(황 씨)) — Its gripped rim is turned partly toward the camera while the central assembly recedes into the wheelhouse; used as Held beneath 황봉업(황 씨)'s hands as the immediate evidence of control over the boat; 어선 뱃머리 (Advancing through the sea) — Its forward-pointing side recedes away from the camera toward the sea; used as Extends through the background as the endpoint of 황봉업(황 씨)'s hostile sightline; 바다의 안개 (Dense enough to obscure distant visibility); used as Reduces background clarity around the bow and prepares the following reveal.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S54sh3__bgfirst_bg.png",
  "asset_id": "1d80f3d3-d19d-47a7-bd56-87bf701ceada",
  "input_asset_ids": [
   "0e14280b-bc1a-4ad8-8f74-3845c900d7bf",
   "81ef185c-4c25-41b8-ad18-10745e6f506f"
  ]
 },
 "S54sh3": {
  "input_fingerprint": "f125bda0ccd3c9de",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, dense fog.\n\nSHOT TEXT (authoritative, Korean): 조타실 안에서 핸들을 쥔 황봉업(황 씨)이 뱃머리의 정인우를 향해 적대감 가득한 눈빛을 쏘아보는 측면.\n\nLOCATION (lock): Inside the wheelhouse of a fishing boat moving through dense daytime fog. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 황봉업(황 씨) (한국인 50대 중반 남성, 넓고 둥근 얼굴, 짧은 검은 머리, 넉넉한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, dense fog.\n\nSHOT TEXT (authoritative, Korean): 조타실 안에서 핸들을 쥔 황봉업(황 씨)이 뱃머리의 정인우를 향해 적대감 가득한 눈빛을 쏘아보는 측면.\n\nLOCATION (lock): Inside the wheelhouse of a fishing boat moving through dense daytime fog. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 황봉업(황 씨) (한국인 50대 중반 남성, 넓고 둥근 얼굴, 짧은 검은 머리, 넉넉한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, dense fog.\n\nSHOT TEXT (authoritative, Korean): 조타실 안에서 핸들을 쥔 황봉업(황 씨)이 뱃머리의 정인우를 향해 적대감 가득한 눈빛을 쏘아보는 측면.\n\nLOCATION (lock): Inside the wheelhouse of a fishing boat moving through dense daytime fog. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 황봉업(황 씨) (한국인 50대 중반 남성, 넓고 둥근 얼굴, 짧은 검은 머리, 넉넉한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S54sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S54sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 황봉업(황 씨): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:774513>"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_mainland_harbor_dock_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 황봉업(황 씨): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:774513>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "조타실 내부의 측면 구도, 인물의 적대적인 시선, 그리고 뱃머리에 선 인물의 팔 깁스 상태까지 지시사항을 정확하게 반영했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "조타실 내부와 안개 낀 분위기는 잘 표현되었으나, 뱃머리 인물의 깁스가 보이지 않으며 표정의 적대감이 다소 부족합니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "조타실 내부의 측면 구도, 인물의 적대적인 시선, 그리고 뱃머리에 선 인물의 팔 깁스 상태까지 지시사항을 정확하게 반영했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "조타실 내부와 안개 낀 분위기는 잘 표현되었으나, 뱃머리 인물의 깁스가 보이지 않으며 표정의 적대감이 다소 부족합니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지시된 안개 낀 바다와 조타실 배경을 잘 구현했으며, 뱃머리에 있는 인물의 깁스 상태까지 정확하게 표현하여 프롬프트 충실도가 높습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "전반적인 구도와 인물의 표정은 좋으나, 뱃머리에 있는 인물이 실루엣으로만 처리되어 명시된 깁스(cast) 상태가 확인되지 않습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지시된 안개 낀 바다와 조타실 배경을 잘 구현했으며, 뱃머리에 있는 인물의 깁스 상태까지 정확하게 표현하여 프롬프트 충실도가 높습니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "전반적인 구도와 인물의 표정은 좋으나, 뱃머리에 있는 인물이 실루엣으로만 처리되어 명시된 깁스(cast) 상태가 확인되지 않습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 11
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 16,
   "B": 11
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "조타실 내부의 측면 구도, 인물의 적대적인 시선, 그리고 뱃머리에 선 인물의 팔 깁스 상태까지 지시사항을 정확하게 반영했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "조타실 내부와 안개 낀 분위기는 잘 표현되었으나, 뱃머리 인물의 깁스가 보이지 않으며 표정의 적대감이 다소 부족합니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_mainland_harbor_dock_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 황봉업(황 씨): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:774513>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "샷 텍스트에 명시된 '적대감 가득한 눈빛'이 드러나지 않고 비교적 평온한 무표정입니다.",
     "fix_en": "Modify Mr. Hwang's facial expression to show a hostile, intense glare with furrowed brows.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "레이아웃 스케치에서는 왼손이 조타 핸들의 우측 상단을 잡고 있으나, 생성된 이미지에서는 좌측 상단을 잡고 있어 포즈가 어긋납니다.",
     "fix_en": "Move the character's left hand to grip the upper-right side of the steering wheel, matching the pose in the layout sketch.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "원본 배경의 고정 요소인 조타 핸들의 형태(바퀴살 5개)가 임의로 바퀴살 4개로 변형되었습니다.",
     "fix_en": "Restore the steering wheel to exactly match the 5-spoke design and center hub of the original background reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Modify Mr. Hwang's facial expression to show a hostile, intense glare with furrowed brows.\n- Move the character's left hand to grip the upper-right side of the steering wheel, matching the pose in the layout sketch.\n- Restore the steering wheel to exactly match the 5-spoke design and center hub of the original background reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S54sh3__bgfirst_bg.png",
   "bg_asset_id": "1d80f3d3-d19d-47a7-bd56-87bf701ceada",
   "bg_record_key": "S54sh3::bgfirst_bg",
   "chain_winner": true,
   "authority": "seed_bg"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S54sh4": {
  "input_fingerprint": "25d41352d403e129",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, dense fog.\n\nSHOT TEXT (authoritative, Korean): 짙은 안개 사이로 어둡고 음울한 기운을 뿜어내는 금월도의 기괴한 형상이 우뚝 솟아 있는 전경.\n\nLOCATION (lock): Across the fog-covered sea ahead of the fishing boat, where a dark island emerges from the mist. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 5,
   "C": 7
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "어선 앞이라는 카메라 시점을 올바르게 해석하여 배를 배제했으며, 짙은 안개 속 기괴하고 음울한 섬의 형상을 가장 훌륭하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "어선의 앞쪽 바다를 비춰야 하는 시점임에도 어선 자체를 프레임 안에 포함시켜 프롬프트의 의도된 구도를 벗어났습니다."
   },
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "지정된 카메라 시점을 잘 따르고 사실적인 파도를 묘사했으나, 섬의 형태나 분위기 측면에서 기괴하고 음울한 느낌이 A에 비해 부족합니다."
   }
  ],
  "refs": [],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S55sh3::bgfirst_bg": {
  "input_fingerprint": "7e5f29c333f0bc4e",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 선착장에 모인 금월도 마을 주민들이 정인우를 향해 적대적이고 차가운 시선을 던지는 광경.\n\nLOCATION (lock): On the island pier beside the newly arrived fishing boat and hostile villagers.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height just behind and slightly beside 정인우, pan across a slightly wide arrangement of several 금월도 마을 주민 while retaining his hesitant shoulder and torso at the near edge. Distribute the residents across the middle distance in staggered depth, allowing their converging hostile gazes—not a frontal lineup—to make 정인우 feel surrounded.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 정인우 in the lower-left of the frame, foreground, looks toward gathered villagers; 금월도 마을 주민 in the middle-center of the frame, midground, looks toward Jeong Inwoo.\n- KEY BACKGROUND ELEMENTS: 금월도 선착장 (Occupied by 정인우 and gathered villagers); used as Provides the open shared space between 정인우 and the gathered residents; 정박한 어선 (At the dock after arrival) — The side facing the dock is visible behind 정인우's edge position; used as Marks the direction from which 정인우 has entered the village.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 선착장에 모인 금월도 마을 주민들이 정인우를 향해 적대적이고 차가운 시선을 던지는 광경.\n\nLOCATION (lock): On the island pier beside the newly arrived fishing boat and hostile villagers.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height just behind and slightly beside 정인우, pan across a slightly wide arrangement of several 금월도 마을 주민 while retaining his hesitant shoulder and torso at the near edge. Distribute the residents across the middle distance in staggered depth, allowing their converging hostile gazes—not a frontal lineup—to make 정인우 feel surrounded.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 정인우 in the lower-left of the frame, foreground, looks toward gathered villagers; 금월도 마을 주민 in the middle-center of the frame, midground, looks toward Jeong Inwoo.\n- KEY BACKGROUND ELEMENTS: 금월도 선착장 (Occupied by 정인우 and gathered villagers); used as Provides the open shared space between 정인우 and the gathered residents; 정박한 어선 (At the dock after arrival) — The side facing the dock is visible behind 정인우's edge position; used as Marks the direction from which 정인우 has entered the village.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S55sh3__bgfirst_bg.png",
  "asset_id": "f3588c91-2ba2-49b9-a88d-3ef43606a76c",
  "input_asset_ids": [
   "6622e07f-e320-4411-b8e2-a2fc157634e4",
   "a553bc1d-51b8-4416-9cd3-f844bfb2f2b9",
   "bac2f48d-abbf-4aa6-94c9-42103de83af2"
  ]
 },
 "S55sh3": {
  "input_fingerprint": "5b8c5ad49c76ed83",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 선착장에 모인 금월도 마을 주민들이 정인우를 향해 적대적이고 차가운 시선을 던지는 광경.\n\nLOCATION (lock): On the island pier beside the newly arrived fishing boat and hostile villagers. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 선착장에 모인 금월도 마을 주민들이 정인우를 향해 적대적이고 차가운 시선을 던지는 광경.\n\nLOCATION (lock): On the island pier beside the newly arrived fishing boat and hostile villagers. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 선착장에 모인 금월도 마을 주민들이 정인우를 향해 적대적이고 차가운 시선을 던지는 광경.\n\nLOCATION (lock): On the island pier beside the newly arrived fishing boat and hostile villagers. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S55sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S55sh3.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L15B02.png"
    },
    {
     "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_fishing_boat_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "인물들 사이에 흰색 화살표와 선(마커/다이어그램)이 렌더링된 하드 위반으로 인해 실격입니다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "레퍼런스 사진의 장소와 구조물을 정확히 반영했으며, 깁스를 한 정인우와 적대적인 주민들의 구도를 성공적으로 연출했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "인물들 사이에 흰색 화살표와 선(마커/다이어그램)이 렌더링된 하드 위반으로 인해 실격입니다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "레퍼런스 사진의 장소와 구조물을 정확히 반영했으며, 깁스를 한 정인우와 적대적인 주민들의 구도를 성공적으로 연출했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 로케이션과 인물의 깁스 상태를 정확히 반영했으며, 마을 주민들의 적대적인 시선이 자연스럽게 연출되었습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "주민들의 시선을 나타내는 흰색 화살표(마커/다이어그램)가 화면에 그대로 노출되어 심각한 규정 위반(Hard Violation)입니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 로케이션과 인물의 깁스 상태를 정확히 반영했으며, 마을 주민들의 적대적인 시선이 자연스럽게 연출되었습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "주민들의 시선을 나타내는 흰색 화살표(마커/다이어그램)가 화면에 그대로 노출되어 심각한 규정 위반(Hard Violation)입니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 5,
     "B": 15
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 5,
   "B": 15
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 2,
    "verdict_ko": "인물들 사이에 흰색 화살표와 선(마커/다이어그램)이 렌더링된 하드 위반으로 인해 실격입니다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "레퍼런스 사진의 장소와 구조물을 정확히 반영했으며, 깁스를 한 정인우와 적대적인 주민들의 구도를 성공적으로 연출했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L15B02.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_fishing_boat_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "정인우의 배 앞부분, 파란색 팔걸이(슬링) 상단 가장자리에 해부학적으로 불가능한 세 번째 손이 얹혀져 있습니다.",
     "fix_en": "Remove the extra, unattached human hand resting on the top edge of the blue sling against Inwoo's stomach.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "보트 뒤편 배경이 바다와 먼 섬으로 이루어진 장소 참고 사진과 달리, 흰색 건물과 마을이 잘못 배치되어 주변 환경의 일관성을 위반했습니다.",
     "fix_en": "Replace the white building and village background behind the boat with open water and distant hills to match the exact surroundings of the location photograph.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the extra, unattached human hand resting on the top edge of the blue sling against Inwoo's stomach.\n- Replace the white building and village background behind the boat with open water and distant hills to match the exact surroundings of the location photograph.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S55sh3__bgfirst_bg.png",
   "bg_asset_id": "f3588c91-2ba2-49b9-a88d-3ef43606a76c",
   "bg_record_key": "S55sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate",
   "seed_attached": true
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S55sh5": {
  "input_fingerprint": "dffe5d9e78f2ee9b",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 반월 식당 앞에 우두커니 선 강영란(흰 머리 여자)이 선착장 쪽을 뚫어지게 주시하는 모습.\n\nLOCATION (lock): Directly outside the island restaurant, overlooking the pier and village approach. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the island pier, hostile villagers, moored boat, restaurant frontage, and dull daylight. Exclude the arriving man's crowd-centered framing and isolate the elder before the restaurant.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 5,
   "A": 2
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "마을 사람들이 화면에서 누락된 점은 아쉬우나, 식당 앞에 홀로 선 흰머리 노인을 고립시키라는 샷 구도와 프레이밍 지시를 가장 충실하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "지정된 마을 사람들을 유지하고 도착한 남자를 제외하는 데는 성공했으나, 노인을 고립시키지 않고 군중 속에 그대로 배치하여 프레이밍 지시를 어겼습니다."
   },
   {
    "label": "A",
    "score": 2,
    "verdict_ko": "제외해야 할 남자를 화면에 남겨둔 채 깁스 위치까지 반전시켰으며, 빈 간판에 금지된 텍스트를 임의로 추가하고 노인 고립 지시도 실패한 심각한 위반입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S55sh3_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "노인 여성이 이전 샷에 없던 어두운 앞치마를 입고 있어 복장 연속성에 어긋납니다.",
     "fix_en": "Remove the dark apron, restoring the dark trousers she wears in the reference image.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "여성의 왼쪽 손(화면 우측)이 해부학적으로 심하게 뭉개져 기형적으로 보입니다.",
     "fix_en": "Reconstruct the woman's left hand (viewer's right) into a fully formed, normal human hand.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "텍스트 생성을 금지했으나 보트 좌측 후미에 글자가 적혀 있습니다.",
     "fix_en": "Erase the painted text from the back left hull of the boat.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the dark apron, restoring the dark trousers she wears in the reference image.\n- Reconstruct the woman's left hand (viewer's right) into a fully formed, normal human hand.\n- Erase the painted text from the back left hull of the boat.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S55sh3"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S55sh7": {
  "input_fingerprint": "103355567d3b30aa",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)의 등 뒤에서 정인우를 향해 한 팔을 높이 들어 올린 채 반갑게 손인사 자세를 취하고 있는 김지후(강수리영)의 모습.\n\nLOCATION (lock): At the restaurant entrance behind the older woman, facing the island pier. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the pier, village crowd, restaurant frontage, and overcast daylight. Exclude the elder's solitary pose and add the younger woman waving from behind her.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains missing, and Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 5,
   "A": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "지정된 카메라 구도(식당 입구, 노인의 등 뒤)와 부두를 향한 시점, 손을 흔드는 인물의 동작을 가장 정확하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "카메라 위치는 식당 쪽이나 노인의 등 뒤가 아닌 측면을 보여주며, 금지된 텍스트가 유리창에 생성되어 감점되었습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "지정된 카메라 위치(식당 입구)와 방향(부두를 향함)을 완전히 위반하고 부두에서 식당을 바라보는 역방향으로 연출된 치명적 오류가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S55sh5_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "김지후가 들고 있는 오른손의 손바닥이 앞을 향할 때 엄지손가락이 안쪽(왼쪽)에 있어야 하나, 바깥쪽(오른쪽)에 달려 있어 해부학적으로 틀렸습니다.",
     "fix_en": "Redraw the raised right hand to have correct anatomy, moving the thumb to the left side so it reads as a right hand with the palm facing forward.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "레퍼런스 이미지에서 노인 여성의 머리는 묶지 않은 짧은 단발이지만, 생성된 이미지에서는 머리가 뒤로 묶여 있어 일관성이 없습니다.",
     "fix_en": "Change the older woman's hair to be short and worn loose, removing the tied ponytail to match the hairstyle established in the reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the raised right hand to have correct anatomy, moving the thumb to the left side so it reads as a right hand with the palm facing forward.\n- Change the older woman's hair to be short and worn loose, removing the tied ponytail to match the hairstyle established in the reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S55sh5"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "groupbg::바위해안_길": {
  "input_fingerprint": "666ed385d06f0856",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "바위해안_길",
    "tags": [
     "S56sh3",
     "S56sh6",
     "S56sh7",
     "S60sh6",
     "S60sh9"
    ]
   },
   "context_sig": "8f67492786de60fd"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On a narrow island coastal path between the village and the rocky shoreline.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인) / 숲속 도축장 내부: 벽에 도축용 기구들이 널려 있는 가건물 내부. 바닥과 벽면에 검붉은 핏자국이 선명하다. (특징: 도축용 칼과 뾰족한 금속 갈고리들; 피받이 양동이; 핏자국이 흥건한 도축대, 바닥, 벽면; 구석에 쌓인 피 묻은 포대자루 더미; 포대자루에서 쏟아진 토막 난 인체 부위)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 해안길 - 낮\n- 길이 사라지며 가파른 바위들이 이어진다.\n- 실외. 해안길 - 낮\n- 다시 해안가 바위길로 나온 수리영과 인우\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On a narrow island coastal path between the village and the rocky shoreline.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인) / 숲속 도축장 내부: 벽에 도축용 기구들이 널려 있는 가건물 내부. 바닥과 벽면에 검붉은 핏자국이 선명하다. (특징: 도축용 칼과 뾰족한 금속 갈고리들; 피받이 양동이; 핏자국이 흥건한 도축대, 바닥, 벽면; 구석에 쌓인 피 묻은 포대자루 더미; 포대자루에서 쏟아진 토막 난 인체 부위)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 해안길 - 낮\n- 길이 사라지며 가파른 바위들이 이어진다.\n- 실외. 해안길 - 낮\n- 다시 해안가 바위길로 나온 수리영과 인우\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_바위해안_길_22ad18.png",
  "asset_id": "dc117794-c96b-4f83-a7bd-f37661f38cb0",
  "input_asset_ids": [
   "478adb86-aa03-4e21-8c6f-476e792234a5"
  ],
  "origin_tag": "S56sh3",
  "place_text": "On a narrow island coastal path between the village and the rocky shoreline.",
  "origin_inputs": {
   "place_text": "On a narrow island coastal path between the village and the rocky shoreline.",
   "time_of_day_en": "day",
   "conti_asset_id": "478adb86-aa03-4e21-8c6f-476e792234a5"
  }
 },
 "S56sh3::bgfirst_bg": {
  "input_fingerprint": "535681c46d3d705c",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 오던 길에 우뚝 선 채 어이없다는 표정으로 김지후(강수리영)를 쏘아보는 정인우의 찰나.\n\nLOCATION (lock): On a narrow island coastal path between the village and the rocky shoreline.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at eye level in front of 정인우 and slightly toward 김지후(강수리영)'s side of the narrow path, framing 정인우 in a medium three-quarter view as his forward movement stops. Keep 김지후(강수리영)'s shoulder and partial profile at the opposite edge so his disbelieving stare has an explicit destination without turning the moment into a symmetrical two-shot.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 좁은 해안길 (Narrow enough for the pair to walk together); used as Compresses the two characters into a confrontational shared axis; 해안과 바다 (Bordering the path); used as Remains visible beyond the path as spatial context for the coastal route.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 오던 길에 우뚝 선 채 어이없다는 표정으로 김지후(강수리영)를 쏘아보는 정인우의 찰나.\n\nLOCATION (lock): On a narrow island coastal path between the village and the rocky shoreline.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at eye level in front of 정인우 and slightly toward 김지후(강수리영)'s side of the narrow path, framing 정인우 in a medium three-quarter view as his forward movement stops. Keep 김지후(강수리영)'s shoulder and partial profile at the opposite edge so his disbelieving stare has an explicit destination without turning the moment into a symmetrical two-shot.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 좁은 해안길 (Narrow enough for the pair to walk together); used as Compresses the two characters into a confrontational shared axis; 해안과 바다 (Bordering the path); used as Remains visible beyond the path as spatial context for the coastal route.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S56sh3__bgfirst_bg.png",
  "asset_id": "d944b2db-cdf0-4eb3-a1b6-a102d1d903ec",
  "input_asset_ids": [
   "478adb86-aa03-4e21-8c6f-476e792234a5",
   "dc117794-c96b-4f83-a7bd-f37661f38cb0"
  ]
 },
 "S56sh3": {
  "input_fingerprint": "b23fb681b21df0c9",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 오던 길에 우뚝 선 채 어이없다는 표정으로 김지후(강수리영)를 쏘아보는 정인우의 찰나.\n\nLOCATION (lock): On a narrow island coastal path between the village and the rocky shoreline. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast; Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 오던 길에 우뚝 선 채 어이없다는 표정으로 김지후(강수리영)를 쏘아보는 정인우의 찰나.\n\nLOCATION (lock): On a narrow island coastal path between the village and the rocky shoreline. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast; Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 오던 길에 우뚝 선 채 어이없다는 표정으로 김지후(강수리영)를 쏘아보는 정인우의 찰나.\n\nLOCATION (lock): On a narrow island coastal path between the village and the rocky shoreline. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast; Jihu’s pendant remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S56sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S56sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:903809>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_바위해안_길_22ad18.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:903809>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "상대방을 오버더숄더로 프레임에 배치하여 '지후를 쏘아보는' 상황을 명확히 구현했으며, 어이없어하는 표정 연기가 지문과 정확히 일치합니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "캐릭터 레퍼런스의 의상(녹색 재킷)은 더 잘 반영했으나, 시선이 화면 밖을 향해 대상과의 상호작용이 약하고 표정도 다소 평이합니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "상대방을 오버더숄더로 프레임에 배치하여 '지후를 쏘아보는' 상황을 명확히 구현했으며, 어이없어하는 표정 연기가 지문과 정확히 일치합니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "캐릭터 레퍼런스의 의상(녹색 재킷)은 더 잘 반영했으나, 시선이 화면 밖을 향해 대상과의 상호작용이 약하고 표정도 다소 평이합니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 인물(정인우)만 프레임에 담아냈으며, 참조 이미지와 유사한 의상(녹색 재킷)과 깁스 설정, 쏘아보는 표정을 잘 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "허용된 인물 목록(PEOPLE)에 없는 인물이 프레임에 등장(오버더숄더)하여 지침을 위반했으며, 정인우의 의상도 참조 이미지와 다릅니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 인물(정인우)만 프레임에 담아냈으며, 참조 이미지와 유사한 의상(녹색 재킷)과 깁스 설정, 쏘아보는 표정을 잘 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "허용된 인물 목록(PEOPLE)에 없는 인물이 프레임에 등장(오버더숄더)하여 지침을 위반했으며, 정인우의 의상도 참조 이미지와 다릅니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 13
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 11,
   "B": 13
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "상대방을 오버더숄더로 프레임에 배치하여 '지후를 쏘아보는' 상황을 명확히 구현했으며, 어이없어하는 표정 연기가 지문과 정확히 일치합니다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "캐릭터 레퍼런스의 의상(녹색 재킷)은 더 잘 반영했으나, 시선이 화면 밖을 향해 대상과의 상호작용이 약하고 표정도 다소 평이합니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_바위해안_길_22ad18.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903809>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "재킷의 왼쪽 가슴과 왼쪽 소매에 영문 텍스트(로고)가 생성되어 '텍스트 없음' 지침을 위반했습니다.",
     "fix_en": "Remove all text and logos from the jacket's chest and sleeve, making the fabric completely plain.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all text and logos from the jacket's chest and sleeve, making the fabric completely plain.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S56sh3__bgfirst_bg.png",
   "bg_asset_id": "d944b2db-cdf0-4eb3-a1b6-a102d1d903ec",
   "bg_record_key": "S56sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "바위해안_길",
   "groupbg_asset_id": "dc117794-c96b-4f83-a7bd-f37661f38cb0"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S56sh6": {
  "input_fingerprint": "88cbede5f7fe300b",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 가파른 해안가 바위벽에 찰싹 달라붙어 조심스럽게 위쪽 바위를 향해 한쪽 발을 뻗은 김지후(강수리영)와 정인우의 모습.\n\nLOCATION (lock): On a steep, exposed rock face where the coastal path disappears above the sea. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains missing, and Inwoo’s injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상); 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 5,
   "C": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "바위벽에 찰싹 달라붙어 발을 뻗은 지후의 묘사가 정확하며, 인우의 깁스와 지후의 목걸이 부재 등 상태가 잘 유지되었습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "지후가 바위에 밀착된 느낌이 부족하며, 인물 간의 크기 비율이 어색하고 행동이 다소 부자연스럽습니다."
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "바위를 오르는 자세는 잘 표현되었으나, 인우의 깁스 끝부분이 손이 아닌 원통형 물체처럼 왜곡되어 렌더링되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S56sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903809>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "바위를 짚고 있는 김지후의 왼손 손가락 형태가 뭉개져 있고 비정상적으로 렌더링되었습니다.",
     "fix_en": "Redraw Jihu's left hand to have a normal human anatomy with five distinct fingers properly gripping the rock.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "이전 숏 레퍼런스에서 왼쪽 어깨로 대각선으로 넘어가던 정인우의 팔걸이 끈이, 오른쪽 어깨 위로 곧장 올라가게 잘못 그려졌습니다.",
     "fix_en": "Remove the sling strap going straight up Inwoo's right shoulder and redraw it crossing diagonally over his chest towards his left shoulder.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "정인우의 깁스 밖으로 노출된 오른손이 손가락 구분이 없는 뭉툭한 형태로 그려졌습니다.",
     "fix_en": "Define clear, anatomically correct fingers emerging from the white bandage on Inwoo's right hand.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw Jihu's left hand to have a normal human anatomy with five distinct fingers properly gripping the rock.\n- Remove the sling strap going straight up Inwoo's right shoulder and redraw it crossing diagonally over his chest towards his left shoulder.\n- Define clear, anatomically correct fingers emerging from the white bandage on Inwoo's right hand.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S56sh3"
  }
 },
 "S56sh7": {
  "input_fingerprint": "866aad1ddc6697a6",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 멀리 떨어진 바다 위 어선에서 바위를 오르는 두 사람을 뚫어지게 감시하는 황봉업(황 씨)의 측면.\n\nLOCATION (lock): Aboard a fishing boat offshore from the rocky island coast, looking toward the climbers. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the steep coastal rocks, narrow shoreline route, sea, and daylight. Exclude the close climbing emphasis and show the distant fisherman watching from his boat.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 황봉업(황 씨) (한국인 50대 중반 남성, 넓고 둥근 얼굴, 짧은 검은 머리, 넉넉한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "캐릭터 레퍼런스의 얼굴과 복장을 정확히 재현했으며, 두 사람을 감시하는 측면 구도를 훌륭하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "지정된 인물의 외모와 복장이 일치하지 않으며, 절벽 위 인물들의 디테일이 이전 샷과 다릅니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "절벽을 오르는 사람이 한 명뿐이라 '두 사람'이라는 지시를 위반했고, 인물의 외모도 완전히 다릅니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S56sh6_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 황봉업(황 씨): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:774513>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트에 지시되지 않은 망원경이 임의로 생성되었으며, 심지어 눈이 아닌 이마에 잘못 대고 있습니다.",
     "fix_en": "Remove the telescope and the hand holding it entirely, restoring the man's bare face and the boat window behind it.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the telescope and the hand holding it entirely, restoring the man's bare face and the boat window behind it.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S56sh6"
  }
 },
 "S57sh2": {
  "input_fingerprint": "2fde8f1cbee9d337",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 정인우의 시선 끝에 놓인 낡은 비석 표면에 '수 리 영' 세 글자가 희미하게 새겨진 찰나의 클로즈업.\n\nLOCATION (lock): At a weathered marker stone beside a forest route reached from the coastal path. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The marker remains inscribed “Su-ri-yeong,” with rotten meat left before it. Jihu’s pendant remains missing.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 8,
   "A": 6,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "로케이션의 비석 질감과 구조를 잘 유지하며, 지시된 세 글자와 고기 소품을 정확한 클로즈업으로 담아냈습니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "클로즈업 연출은 무난하나, 프롬프트에서 요구하지 않은 영문 텍스트가 불필요하게 노출되었습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "레퍼런스의 길쭉한 비석 형태를 무시하고 짧고 둥근 형태로 변형하여 로케이션 고정 조건을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_coastal_path_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "비석에 새겨진 '수 리 영' 세 글자가 너무 선명하고 깊게 파여 있습니다. (프롬프트 요구사항: 희미하게 새겨진)",
     "fix_en": "Make the engraved Korean characters '수 리 영' much shallower, faded, and fainter so they barely stand out from the weathered stone surface.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "비석 단상에 놓인 고기가 썩은 고기가 아니라 조리된 구운 고기처럼 보입니다.",
     "fix_en": "Change the pile of cooked-looking meat on the stone ledge into a lump of raw, discolored, and rotting meat.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Make the engraved Korean characters '수 리 영' much shallower, faded, and fainter so they barely stand out from the weathered stone surface.\n- Change the pile of cooked-looking meat on the stone ledge into a lump of raw, discolored, and rotting meat.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "seed-bg만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_bypass:bg_only"
 },
 "S57sh3": {
  "input_fingerprint": "bbf83926243ec5b1",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 비석 앞의 정인우 등 뒤에 선 채 미간을 잔뜩 찌푸린 김지후(강수리영)의 얼굴.\n\nLOCATION (lock): Immediately behind the man examining the marker stone in the daytime forest. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the mossy stone, rotten meat offering, dense foliage, and daylight. Exclude the inscription-only framing and show the woman standing behind the inspecting man.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains missing, and Inwoo’s injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "정인우의 등 뒤에 선 위치와 레퍼런스의 배경, 고기 공물을 올바르게 구현했으나 남자의 팔 깁스가 누락되었습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "3순위 인물 배치 위반; 여자가 남자의 등 뒤가 아닌 비석 뒤편에 서 있습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "3순위 장소 위반; 배경 계절이 다르고 지시된 고기 공물이 누락되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S57sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "비석 표면의 글씨가 이전 샷에 있던 '수리영'이 아닌 알아볼 수 없는 형태로 왜곡되어 있습니다.",
     "fix_en": "Restore the clear Korean inscription '수리영' on the tombstone, matching the reference image's engraving.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "정인우의 부상당한 팔에 단단한 깁스(cast)가 아닌 파란색 천 소재의 팔걸이(sling)가 둘러져 있습니다.",
     "fix_en": "Change the blue fabric sling on the man's arm into a rigid medical plaster cast.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore the clear Korean inscription '수리영' on the tombstone, matching the reference image's engraving.\n- Change the blue fabric sling on the man's arm into a rigid medical plaster cast.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S57sh2"
  }
 },
 "S57sh5": {
  "input_fingerprint": "b04ca22d907df474",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 바짝 긴장한 얼굴로 날카롭게 눈을 빛내며 어두운 수풀을 팔로 밀어내고 한 발을 앞으로 뻗어 내딛은 미드액션 자세의 김지후(강수리영) 전신.\n\nLOCATION (lock): Along an overgrown forest route beyond the name-carved marker stone. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the dense forest, marker stone area, dark undergrowth, and filtered daylight. Exclude the stationary inspection pose and show the woman pushing forward through brush.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains missing, and Inwoo’s injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 4,
   "B": 7,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "여성의 행동은 잘 표현했으나, 지문에 없는 남성 캐릭터가 이전 샷과 다른 자세로 등장하여 인물 및 연속성 규칙을 위반했습니다."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지문이 요구한 전신 프레이밍과 수풀을 헤치는 동작을 정확히 연출했으며, 지문에 없는 인물을 배제하라는 제한 규칙을 유일하게 준수했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "등장하지 않아야 할 남성이 묘비에 기댄 전혀 다른 자세로 렌더링되어, 인물 제한 및 연속성 규칙을 심각하게 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S57sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "배경에 있는 비석이 참조 이미지에 설정된 길고 평평하며 이름이 각인된 비석과 전혀 다르게, 이끼가 낀 형태의 사각 기둥으로 잘못 묘사되었습니다.",
     "fix_en": "Change the mossy stone structure in the background on the left to match the tall, flat, light-colored marker stone with engraved Korean text from the reference image.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "프롬프트에서 전신(전신) 샷을 요구했으나, 프레임 하단이 캐릭터의 발목 부근에서 잘려 발이 온전히 보이지 않습니다.",
     "fix_en": "The camera framing cuts off the character's feet, failing the requirement for a full-body shot.",
     "needs_regeneration": false,
     "unfixable": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the mossy stone structure in the background on the left to match the tall, flat, light-colored marker stone with engraved Korean text from the reference image.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S57sh3"
  }
 },
 "S58sh1": {
  "input_fingerprint": "2da06db8205d1ccd",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 검은 염소 떼가 사라지고 텅 비어 있는 낡고 어두운 축사의 적막한 내부 전경.\n\nLOCATION (lock): Inside the dark, abandoned-looking goat shed, along the aisle through empty pens. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The goat enclosure remains empty; Jihu’s pendant remains missing and Inwoo’s injured arm remains cast.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 6,
   "C": 2
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지시문에 맞게 낡고 어두운 빈 축사의 적막한 내부를 금지된 텍스트 없이 잘 구현했습니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "빈 축사의 분위기를 잘 표현했으나 구조물과 창문 형태가 다소 부자연스럽습니다."
   },
   {
    "label": "C",
    "score": 2,
    "verdict_ko": "이미지 하단에 프롬프트에서 명시적으로 금지한 텍스트(자막)가 출력되어 심각한 규칙 위반입니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "생성된 이미지는 장소의 절대적 기준인 레퍼런스 사진(비석이 있는 야외 숲속 오솔길)을 완전히 무시하고 축사 내부 공간을 생성했습니다.",
     "fix_en": "Replace the entire indoor barn environment with the outdoor forest path and stone monuments shown in the reference photograph.",
     "needs_regeneration": true,
     "unfixable": true
    }
   ]
  },
  "fix_skipped": true,
  "fix_skip_reason": "all_issues_unfixable",
  "ref_mode": "seed-bg만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_bypass:bg_only"
 },
 "S58sh4": {
  "input_fingerprint": "187e40b295e5871c",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 극도의 긴장감에 휩싸여 가쁜 숨을 들이마신 채 굳어 있는 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): Within the empty goat shed along the same narrow passage used the previous night. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the empty wooden goat enclosure, dark stalls, rough fencing, and filtered daylight. Exclude the empty establishing emphasis and frame the woman's tense reaction inside it.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains missing, and Inwoo’s injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 3,
   "B": 8,
   "C": 5
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "긴장한 표정은 잘 표현되었으나, 레퍼런스와 완전히 다른 흰색 의상을 입고 있어 복장 일치도가 매우 낮습니다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "가쁜 숨을 들이마시는 긴장된 표정을 훌륭하게 연출했으며, 지정된 의상(펜던트 없음)과 배경을 정확히 반영했습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "의상과 배경 설정은 정확하나, 가쁜 숨을 쉬는 극도의 긴장감 대신 다소 평온하게 입을 다물고 있어 표정 연출이 아쉽습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S58sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S58sh1"
  }
 },
 "S59sh6": {
  "input_fingerprint": "e65b174c7d26397d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 풀어진 포댓자루 틈새 밖으로 피범벅이 된 창백한 사람 팔 하나가 삐져나와 있는 광경.\n\nLOCATION (lock): Inside the small storage compartment attached to the forest slaughterhouse. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The dismembered remains are packed inside blood-soaked sacks, with one pale severed arm protruding through an opened sack. Jihu’s pendant remains missing and Inwoo’s arm remains cast.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 5,
   "B": 4,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "레퍼런스의 장소와 구도는 훌륭하게 재현했으나, 지시문인 '피범벅이 된 창백한 사람 팔' 대신 장갑 낀 팔이 등장한 점이 아쉽습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "공간의 디테일(문틀 등)이 일부 왜곡되었으며, 역시 피와 창백한 맨살 대신 소매와 장갑이 묘사되었습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "화면 좌측에 영화 촬영용 카메라가 그대로 노출되는 치명적인 요소 개입(Hard Violation)이 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L25B01.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "포대 밖으로 삐져나온 팔이 소매가 있는 옷을 입고 작업용 장갑을 끼고 있습니다. 지문대로 피범벅이 된 창백한 맨살의 팔이어야 합니다.",
     "fix_en": "Change the sleeved and gloved arm protruding from the sack into a bare, pale human arm heavily covered in blood.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "포댓자루들이 피에 젖어있지 않고 그저 더러울 뿐입니다. 피범벅이 된 포댓자루라는 설정에 어긋납니다.",
     "fix_en": "Add dark red, heavy blood stains soaking the fabric of the burlap sacks.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "오른쪽 벽에 레퍼런스 공간의 고정 구조물에 없는 높은 나무 선반이 임의로 추가되었습니다.",
     "fix_en": "Remove the tall wooden shelving unit on the right side of the room, replacing it with the plain concrete wall seen in the reference location.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the sleeved and gloved arm protruding from the sack into a bare, pale human arm heavily covered in blood.\n- Add dark red, heavy blood stains soaking the fabric of the burlap sacks.\n- Remove the tall wooden shelving unit on the right side of the room, replacing it with the plain concrete wall seen in the reference location.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S59sh12": {
  "input_fingerprint": "a936742e53b4f854",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 활짝 열린 창고 문턱에 고산들과 금월도 마을 청년들이 우뚝 선 전경.\n\nLOCATION (lock): At the open threshold between the slaughterhouse and its sack-filled storage compartment. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the bloodstained slaughterhouse storeroom, stacked sacks, open doorway, and daylight leaking in. Exclude the protruding arm emphasis and show the man and youths at the threshold.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The storage room remains stacked with sacks full of dismembered human bodies and seeped blood. Jihu’s pendant remains missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "지시된 전경(wide shot) 구도를 잘 구현했으며, 고정되어야 할 시신의 팔을 유지하면서도 샷의 초점을 인물들로 정확히 이동시킴."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "시신의 팔 위치와 인물 배치는 양호하나, 프롬프트에 없는 무기(쇠스랑, 쇠지렛대)를 임의로 생성하여 감점됨."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "카메라 구도만 바뀌어야 함에도 이전 샷에서 고정 지시를 받은 시신의 팔을 완전히 삭제하여 엄격한 록(lock) 조건을 위반함."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S59sh6_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 컷의 '돌출된 팔 강조를 제외하라'는 지시에도 불구하고, 전경의 자루에서 창백한 팔(또는 다리)이 여전히 두드러지게 튀어나와 있습니다.",
     "fix_en": "Remove the pale limb protruding from the foreground sack entirely, leaving only the bloodstained burlap.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "레퍼런스 이미지의 고정된 배경 요소가 유지되어야 하나, 우측에 있던 단일 문이 벽 중앙의 커다란 양개형(double) 문으로 바뀌어 창고의 건축 구조가 완전히 달라졌습니다.",
     "fix_en": "Restore the doorway to the single-door structure and off-center placement shown in the reference image.",
     "needs_regeneration": true,
     "unfixable": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the pale limb protruding from the foreground sack entirely, leaving only the bloodstained burlap.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S59sh6"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S59sh16": {
  "input_fingerprint": "936085abbd71a097",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 굳게 닫힌 창고 문 안쪽 어둠 속에서 식은땀을 흘리며 두려움에 떠는 김지후(강수리영)와 정인우의 클로즈업.\n\nLOCATION (lock): In the darkness of the closed slaughterhouse storage compartment, beside blood-soaked sacks. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the cramped dark storeroom, closed wooden door, blood-soaked sacks, and narrow light leaks. Exclude the intruders from inside the frame and show the two hidden figures holding still.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu and Inwoo remain hidden beside the sacks of dismembered corpses; her pendant is missing and his injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상); 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 3,
   "C": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "요구된 클로즈업 프레이밍과 인물들의 두려운 표정, 식은땀, 소품 상태(깁스, 목걸이 부재)를 정확히 구현했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "주인을 알 수 없는 깁스 한 팔이 등장하여 인체 해부학적 오류(Hard Violation)가 발생했습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "클로즈업이 아닌 미디엄 샷으로 렌더링되었으며, 정인우의 의상이 레퍼런스와 다르게 묘사되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S59sh12_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903809>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "정인우(오른쪽 남성)가 캐릭터 레퍼런스에서 착용하고 있는 벙거지 모자와 베이지색 다용도 조끼가 누락되었습니다.",
     "fix_en": "Add the bucket hat and beige multi-pocket vest from the character reference to the man on the right.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the bucket hat and beige multi-pocket vest from the character reference to the man on the right.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S59sh12"
  }
 },
 "S60sh6": {
  "input_fingerprint": "b4a8056aeeb68dee",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 갯바위와 바닷물 사이에 사람 시체가 걸린 채, 들이치는 파도에 밀려 한쪽 팔이 허공에 뜬 채 고정된 순간.\n\nLOCATION (lock): Among shoreline rocks below the island coastal path, where a body is caught between stone and surf. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang’s corpse is wedged between the coastal rocks and the seawater, with the torso caught in place while the incoming surf lifts one slack arm into the air.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 8,
   "A": 6,
   "C": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "바위와 바닷물 사이에 시체가 낀 채 파도에 밀려 팔이 들린 순간을 가장 자연스럽고 사실적으로 묘사했습니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "위치와 상황은 일치하나, 파도에 의해 팔이 들렸다기보다 스스로 팔을 뻗고 있는 것처럼 경직되어 보입니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "시체가 바닷물 사이가 아닌 바위 위에 완전히 누워 있으며, 파도와 팔의 상호작용이 전혀 표현되지 않았습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L27B01.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "허공에 뜬 팔과 손가락이 파도에 밀려 떠오른 축 늘어진 시체의 팔이 아니라, 살아있는 사람이 능동적으로 뻗고 있는 것처럼 뻣뻣하게 힘이 들어가 있습니다.",
     "fix_en": "Relax the raised arm and hand entirely so it hangs limp and slack, appearing passively lifted by the crashing wave rather than actively reaching."
    },
    {
     "issue_ko": "위로 뻗은 손의 엄지손가락 위치와 손가락 관절 구조가 뒤틀려 해부학적으로 어색합니다.",
     "fix_en": "Correct the anatomy of the raised hand so the thumb and fingers are in natural positions for a limp, relaxed hand."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Relax the raised arm and hand entirely so it hangs limp and slack, appearing passively lifted by the crashing wave rather than actively reaching.\n- Correct the anatomy of the raised hand so the thumb and fingers are in natural positions for a limp, relaxed hand.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S60sh9": {
  "input_fingerprint": "84c981bac3a32be5",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 핸드폰 카메라를 바다 쪽 시체를 향해 겨눈 채 굳은 표정으로 서 있는 정인우의 측면.\n\nLOCATION (lock): On the exposed rocky coastal path above the body caught in the water. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the coastal rocks, choppy water, lodged corpse, and daylight. Exclude the corpse-only composition and show the man photographing it from the path.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang’s corpse remains lodged among the rocks while Inwoo photographs it; Jihu’s pendant remains missing and Inwoo’s arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 5,
   "A": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "레퍼런스의 시체 자세와 위치를 정확히 유지했으며, 측면에서 핸드폰을 겨누는 구도와 정인우의 깁스 착용 상태를 가장 충실하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "정인우의 깁스는 표현되었으나, 그대로 유지되어야 할 시체의 자세가 레퍼런스와 다르게 변형되어 감점되었습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "프롬프트에 명시된 정인우의 팔 깁스가 누락되었고, 시체의 자세와 위치 또한 레퍼런스와 일치하지 않습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S60sh6_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903809>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "정인우의 캐릭터 레퍼런스에 있는 벙거지 모자와 베이지색 낚시 조끼가 누락되었습니다.",
     "fix_en": "Add the bucket hat and beige multi-pocket fishing vest to the standing man to match his character reference."
    },
    {
     "issue_ko": "시체의 자세가 고정된 레퍼런스와 다릅니다 (오른팔이 앞으로 뻗어 있지 않고 몸통 옆에 붙어 있음).",
     "fix_en": "Reposition the corpse's right arm to extend outward through the water, exactly matching its fixed pose in the previous shot reference."
    },
    {
     "issue_ko": "스마트폰을 쥐고 있는 뒤쪽 손(왼손)의 손가락 형태가 뭉개져 기형적입니다.",
     "fix_en": "Redraw the fingers of the hand partially visible behind the phone to restore correct human anatomy."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the bucket hat and beige multi-pocket fishing vest to the standing man to match his character reference.\n- Reposition the corpse's right arm to extend outward through the water, exactly matching its fixed pose in the previous shot reference.\n- Redraw the fingers of the hand partially visible behind the phone to restore correct human anatomy.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S60sh6"
  }
 },
 "S61sh1": {
  "input_fingerprint": "d0b0883cfd9106a3",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 노을이 지는 금월도 선착장 바닥의 넓적한 비닐 포대 위에 김여상의 시체가 퉁퉁 불어 누워있는 전경.\n\nLOCATION (lock): On the island pier at sunset, where a recovered body lies on a broad plastic sheet. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang’s waterlogged corpse is positioned supine on the spread plastic sheet, still wearing his fishing clothes and bearing the deep claw wound across his chest.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "전경 숏의 구도, 어업용 작업복, 바닥에 깔린 넓은 비닐(방수포) 등 모든 지시사항을 사실적으로 완벽하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "바닥에 깔린 소재가 비닐 포대가 아닌 일반적인 돗자리에 가까워 설정과 어긋납니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "투명한 비닐은 시신 수습 상황에 다소 어색하며, 선착장과 복장의 디테일이 상대적으로 부족합니다."
   }
  ],
  "refs": [],
  "critique": {
   "issues": [
    {
     "issue_ko": "시체가 물에 '퉁퉁 불어' 있다는 묘사와 달리, 얼굴과 피부가 붓기 없이 멀쩡한 상태입니다.",
     "fix_en": "Swell and discolor the corpse's face and visible hands to make the skin look heavily bloated and waterlogged."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Swell and discolor the corpse's face and visible hands to make the skin look heavily bloated and waterlogged.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S61sh12": {
  "input_fingerprint": "f233a6ad87cdaaac",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 구창모가 위압적인 표정으로 정인우의 바로 앞까지 바짝 다가선 전신.\n\nLOCATION (lock): Beside the gathered crowd and covered body on the sunset pier. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the sunset pier, corpse on plastic, gathered villagers, police boat, and warm fading light. Exclude the broad corpse emphasis and show the local officer confronting the visiting man.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang’s corpse remains on the plastic sheet nearby; Jihu’s pendant remains missing and Inwoo’s arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김여상 (한국인 40대 중반 남성, 평범한 중년 얼굴, 짧은 검은 머리, 소박한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "샷 텍스트가 요구한 '전신' 프레이밍을 유일하게 머리부터 발끝까지 정확히 구현했으나, 방문객이 시신과 동일한 의상을 입고 있는 의상 복제 오류가 있습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "'전신' 샷 요구를 위반하고 무릎 위에서 컷아웃되었으며, 방문객 인물에 시신(김여상)의 얼굴 레퍼런스가 잘못 적용되었습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "마찬가지로 '전신' 프레이밍을 위반한 미디엄 샷이며, 경찰관 캐릭터에 시신의 얼굴 레퍼런스가 오적용되어 인물 정체성 오류가 큽니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S61sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김여상: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1438369>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "참조 이미지에서 똑바로 위를 향하고 있던 시신의 머리가 옆으로 돌아가 있어, 고정된 자세를 정확히 유지해야 하는 규칙을 위반했습니다.",
     "fix_en": "Rotate the corpse's head so the face points straight up, matching its exact pose from the reference image.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "샷 텍스트에 명시된 인물 외에는 추가하지 말라는 규칙을 위반하고 화면 양옆에 여러 명의 마을 사람들과 다른 경찰들이 추가되었습니다.",
     "fix_en": "Remove all the extra people standing on the left and right edges, leaving only the two main standing characters and the corpse.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Rotate the corpse's head so the face points straight up, matching its exact pose from the reference image.\n- Remove all the extra people standing on the left and right edges, leaving only the two main standing characters and the corpse.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S61sh1"
  }
 },
 "S61sh16": {
  "input_fingerprint": "bb7d13e4371620fe",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 정인우를 향해 한 발을 딛는 고산들의 가슴팍을 강영란(흰 머리 여자)이 한 손으로 굳건히 막아서며 제지하는 정지 순간.\n\nLOCATION (lock): Among the villagers blocking access to the police boat on the island pier. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the sunset harbor, crowd, corpse area, and police launch. Exclude the officer's close confrontation and show the elder physically restraining the large man.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang’s corpse is being transferred to Changmo’s patrol boat. Jihu’s pendant remains missing and Inwoo’s arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김여상 (한국인 40대 중반 남성, 평범한 중년 얼굴, 짧은 검은 머리, 소박한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 4,
   "A": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "정인우를 향해 다가서는 남자를 흰 머리 여성이 한 손으로 제지하는 텍스트의 핵심 동작(Staging)을 가장 정확하게 구현했습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "지정된 동작은 잘 표현했으나, 바닥에 놓인 시신과 배경에서 운반되는 시신이 동시에 존재하는 중복 생성 오류(Hard Violation)가 있습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "남자가 정인우를 향하지 않고 다른 곳을 보며 걷고 있어, 프롬프트가 요구한 핵심 상호작용 구도를 완전히 벗어났습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S61sh12_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김여상: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1438369>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "흰 머리 여자가 두 손을 모두 사용하여 남자를 제지하고 있습니다 (지문의 '한 손으로'에 위배됨).",
     "fix_en": "Remove the woman's left hand from the large man's arm and leave only her right hand pushing against his chest.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "남자의 가슴을 밀고 있는 여자의 오른손 엄지손가락이 아래를 향하고 있어 해부학적으로 잘못되었습니다 (왼손처럼 보임).",
     "fix_en": "Redraw the woman's right hand on the man's chest so that the thumb points upwards naturally.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the woman's left hand from the large man's arm and leave only her right hand pushing against his chest.\n- Redraw the woman's right hand on the man's chest so that the thumb points upwards naturally.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S61sh12"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "groupbg::경찰정_갑판": {
  "input_fingerprint": "08bd586419eb9155",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "경찰정_갑판",
    "tags": [
     "S62sh4",
     "S62sh9",
     "S87sh15",
     "S87sh19",
     "S87sh23",
     "S88sh6"
    ]
   },
   "context_sig": "86314762c825603b"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On the open deck of a police patrol boat stopped on the dark early-evening sea.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인) / 금월도 해안 마을: 50여 가구가 채 안 되는 섬마을. 비탈을 따라 알록달록한 지붕의 소담스러운 집들이 단풍 든 나무와 어우러져 있다. (특징: 좁고 야트막한 담장의 마을 골목길; 알록달록한 지붕의 낡은 집들; 집 앞 대문에 걸린 새끼줄과 종이 인형; 마당과 담장에 심어진 짙은 녹색 석창포 약초)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 경찰 순찰정 - 저녁\n- 경찰 순찰정 - 계속\n- 실외. 경찰 순찰정 - 잠시 후\n\nTIME OF DAY (lock): early evening.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On the open deck of a police patrol boat stopped on the dark early-evening sea.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인) / 금월도 해안 마을: 50여 가구가 채 안 되는 섬마을. 비탈을 따라 알록달록한 지붕의 소담스러운 집들이 단풍 든 나무와 어우러져 있다. (특징: 좁고 야트막한 담장의 마을 골목길; 알록달록한 지붕의 낡은 집들; 집 앞 대문에 걸린 새끼줄과 종이 인형; 마당과 담장에 심어진 짙은 녹색 석창포 약초)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 경찰 순찰정 - 저녁\n- 경찰 순찰정 - 계속\n- 실외. 경찰 순찰정 - 잠시 후\n\nTIME OF DAY (lock): early evening.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_경찰정_갑판_77dccd.png",
  "asset_id": "00b3f8b7-848a-4470-84a4-6bdc9ea8683d",
  "input_asset_ids": [
   "606b9643-28e3-4bb4-a879-f2d1b1746de7"
  ],
  "origin_tag": "S62sh4",
  "place_text": "On the open deck of a police patrol boat stopped on the dark early-evening sea.",
  "origin_inputs": {
   "place_text": "On the open deck of a police patrol boat stopped on the dark early-evening sea.",
   "time_of_day_en": "early evening",
   "conti_asset_id": "606b9643-28e3-4bb4-a879-f2d1b1746de7"
  }
 },
 "S62sh4::bgfirst_bg": {
  "input_fingerprint": "3705f347bf52a4b1",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 시체에서 빼낸 반지와 젖은 현금을 쥐고 탐욕스러운 미소를 짓는 구창모의 얼굴 클로즈업.\n\nLOCATION (lock): On the open deck of a police patrol boat stopped on the dark early-evening sea.\n\nTIME OF DAY (lock): early evening.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From below 구창모's eye line beside the deck, finish the dolly-in on a close view of his greedy smile while his raised fist enters the lower third. The wet cash and ring remain clearly legible in his hand but occupy less than a quarter of the image, with enough deck and open water retained at the edges to prepare the coming pan away.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 경찰 순찰정 갑판 (시체 수색이 이루어진 갑판); used as A narrow out-of-focus strip beneath 구창모 anchors the close-up aboard the boat; 바다 (어둑해진 초저녁 바다); used as Kept at the outer edge of frame as the direction toward which the next pan will travel; 젖은 현금 (시체에서 꺼내 구창모가 쥔 상태) — 젖은 지폐의 인쇄된 앞면 일부가 카메라를 향한다; used as Held together in the lower third as evidence of the theft; 반지 (시체에서 빼내 구창모가 쥔 상태); used as Placed against the wet cash as a compact secondary focal point.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 시체에서 빼낸 반지와 젖은 현금을 쥐고 탐욕스러운 미소를 짓는 구창모의 얼굴 클로즈업.\n\nLOCATION (lock): On the open deck of a police patrol boat stopped on the dark early-evening sea.\n\nTIME OF DAY (lock): early evening.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From below 구창모's eye line beside the deck, finish the dolly-in on a close view of his greedy smile while his raised fist enters the lower third. The wet cash and ring remain clearly legible in his hand but occupy less than a quarter of the image, with enough deck and open water retained at the edges to prepare the coming pan away.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 경찰 순찰정 갑판 (시체 수색이 이루어진 갑판); used as A narrow out-of-focus strip beneath 구창모 anchors the close-up aboard the boat; 바다 (어둑해진 초저녁 바다); used as Kept at the outer edge of frame as the direction toward which the next pan will travel; 젖은 현금 (시체에서 꺼내 구창모가 쥔 상태) — 젖은 지폐의 인쇄된 앞면 일부가 카메라를 향한다; used as Held together in the lower third as evidence of the theft; 반지 (시체에서 빼내 구창모가 쥔 상태); used as Placed against the wet cash as a compact secondary focal point.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S62sh4__bgfirst_bg.png",
  "asset_id": "d92246ae-7388-4865-914f-2d9735119759",
  "input_asset_ids": [
   "606b9643-28e3-4bb4-a879-f2d1b1746de7",
   "00b3f8b7-848a-4470-84a4-6bdc9ea8683d"
  ]
 },
 "S62sh4": {
  "input_fingerprint": "6b70a2dcaffb7121",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): early evening.\n\nSHOT TEXT (authoritative, Korean): 시체에서 빼낸 반지와 젖은 현금을 쥐고 탐욕스러운 미소를 짓는 구창모의 얼굴 클로즈업.\n\nLOCATION (lock): On the open deck of a police patrol boat stopped on the dark early-evening sea. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang’s corpse remains on the patrol boat while Changmo strips valuables from it.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (한국인 40대 중후반 남성, 호리한 얼굴, 짧은 검은 머리, 가무잡잡한 피부) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): early evening.\n\nSHOT TEXT (authoritative, Korean): 시체에서 빼낸 반지와 젖은 현금을 쥐고 탐욕스러운 미소를 짓는 구창모의 얼굴 클로즈업.\n\nLOCATION (lock): On the open deck of a police patrol boat stopped on the dark early-evening sea. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang’s corpse remains on the patrol boat while Changmo strips valuables from it.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (한국인 40대 중후반 남성, 호리한 얼굴, 짧은 검은 머리, 가무잡잡한 피부) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): early evening.\n\nSHOT TEXT (authoritative, Korean): 시체에서 빼낸 반지와 젖은 현금을 쥐고 탐욕스러운 미소를 짓는 구창모의 얼굴 클로즈업.\n\nLOCATION (lock): On the open deck of a police patrol boat stopped on the dark early-evening sea. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang’s corpse remains on the patrol boat while Changmo strips valuables from it.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (한국인 40대 중후반 남성, 호리한 얼굴, 짧은 검은 머리, 가무잡잡한 피부) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S62sh4__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S62sh4.png"
    },
    {
     "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:893415>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_경찰정_갑판_77dccd.png"
    },
    {
     "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:893415>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시된 클로즈업 앵글 속에서 인물의 탐욕스러운 표정과 소품이 잘 표현되었으며, 배경 시체의 스케일도 자연스럽습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "배경에 배치된 시체의 크기가 배의 구조물에 비해 비정상적으로 작게 렌더링되어 물리적으로 불가능한 비율(하드 위반)을 보입니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시된 클로즈업 앵글 속에서 인물의 탐욕스러운 표정과 소품이 잘 표현되었으며, 배경 시체의 스케일도 자연스럽습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "배경에 배치된 시체의 크기가 배의 구조물에 비해 비정상적으로 작게 렌더링되어 물리적으로 불가능한 비율(하드 위반)을 보입니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "얼굴 클로즈업, 탐욕스러운 미소, 젖은 현금과 반지 등 프롬프트의 지시사항을 훌륭하게 구현했으며 인물의 외형도 잘 일치합니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "반지가 손이나 지폐 위에 물리적으로 불가능한 형태로 떠 있는 치명적인 오류(Hard Violation)가 있습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "얼굴 클로즈업, 탐욕스러운 미소, 젖은 현금과 반지 등 프롬프트의 지시사항을 훌륭하게 구현했으며 인물의 외형도 잘 일치합니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "반지가 손이나 지폐 위에 물리적으로 불가능한 형태로 떠 있는 치명적인 오류(Hard Violation)가 있습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 11
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 11,
   "B": 11
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지시된 클로즈업 앵글 속에서 인물의 탐욕스러운 표정과 소품이 잘 표현되었으며, 배경 시체의 스케일도 자연스럽습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "배경에 배치된 시체의 크기가 배의 구조물에 비해 비정상적으로 작게 렌더링되어 물리적으로 불가능한 비율(하드 위반)을 보입니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_경찰정_갑판_77dccd.png"
   },
   {
    "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:893415>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트에 명시된 '젖은 현금'과 달리, 창모가 쥐고 있는 지폐가 전혀 젖어 있지 않고 빳빳하게 건조해 보입니다.",
     "fix_en": "Make the paper money look completely soaked, limp, darker in color, and clinging to his fingers.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "'텍스트 없음(no text)' 규칙을 위반하여 지폐 표면에 숫자와 문자가 선명하게 생성되어 있습니다.",
     "fix_en": "Remove all visible numbers and letters from the paper money, leaving only a generic currency pattern.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Make the paper money look completely soaked, limp, darker in color, and clinging to his fingers.\n- Remove all visible numbers and letters from the paper money, leaving only a generic currency pattern.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S62sh4__bgfirst_bg.png",
   "bg_asset_id": "d92246ae-7388-4865-914f-2d9735119759",
   "bg_record_key": "S62sh4::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "경찰정_갑판",
   "groupbg_asset_id": "00b3f8b7-848a-4470-84a4-6bdc9ea8683d"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S62sh9": {
  "input_fingerprint": "c52d53b28d580161",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): early evening.\n\nSHOT TEXT (authoritative, Korean): 어두운 바다 위 두 배 사이에서 황봉업(황 씨)과 구창모가 의미심장한 시선을 교환하는 찰나.\n\nLOCATION (lock): Between a police patrol boat and an approaching fishing boat on the dark open water. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the darkening sea, patrol boat deck, approaching fishing boat, and early-evening light. Exclude the corpse-looting close-up and show the two men exchanging a look across the boats.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeo-sang’s corpse and the valuables Changmo removed from it remain split between Hwang’s boat and Changmo’s pockets.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (한국인 40대 중후반 남성, 호리한 얼굴, 짧은 검은 머리, 가무잡잡한 피부) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 3,
   "C": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "이전 샷의 시체 자세와 방향을 정확히 유지하고 창모의 주머니에 챙긴 돈까지 구현하여 프롬프트를 가장 충실히 따랐습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "시체의 방향이 반대로 뒤집히고 두 인물의 얼굴이 동일하게 복제되는 심각한 오류가 발생했습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "두 사람의 시선 교환은 표현되었으나 시체의 팔 자세가 기준과 다르게 변형되었고 창모의 전리품이 누락되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S62sh4_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:893415>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "갑판 위 시신의 자세가 기준 이미지와 일치하지 않습니다(머리가 옆으로 돌아가 있고 팔이 몸에서 떨어져 굽어 있음).",
     "fix_en": "Restore the corpse's exact pose from the reference: turn its head to face straight up and straighten the visible arm so it lies flat against the side of the body.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore the corpse's exact pose from the reference: turn its head to face straight up and straighten the visible arm so it lies flat against the side of the body.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S62sh4"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S63sh2::bgfirst_bg": {
  "input_fingerprint": "cc6a6f73b5d2d56b",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 창문 밖 골목에 선 금월도 마을 청년들이 식당 안쪽을 향해 기괴한 웃음을 짓는 찰나.\n\nLOCATION (lock): At the restaurant's lattice window, looking from the lit dining room toward the nighttime lane outside.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Remain just inside the lattice window at standing eye level and narrow the outward medium-long view so the 금월도 마을 청년들 cluster behind the grid near frame center. Their faces appear in separate lattice openings as they lean toward the dining room and answer the interior gaze with conspicuous, unsettling smiles.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 격자무늬 나무 창 in the middle-center of the frame, foreground; 금월도 마을 청년들 in the middle-center of the frame, background, looks toward dining room interior.\n- KEY BACKGROUND ELEMENTS: 격자무늬 나무 창 (식당 안에서 바깥을 내다볼 수 있는 상태) — 창의 실내 쪽 면과 격자가 전경에 보이고, 그 너머 골목의 청년들이 보인다; used as The camera looks through its grid, dividing the watching faces into confined compartments; 식당 밖 골목 (청년들이 식당을 감시하고 있는 상태); used as Provides the exterior depth plane immediately behind the watching youths.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 창문 밖 골목에 선 금월도 마을 청년들이 식당 안쪽을 향해 기괴한 웃음을 짓는 찰나.\n\nLOCATION (lock): At the restaurant's lattice window, looking from the lit dining room toward the nighttime lane outside.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Remain just inside the lattice window at standing eye level and narrow the outward medium-long view so the 금월도 마을 청년들 cluster behind the grid near frame center. Their faces appear in separate lattice openings as they lean toward the dining room and answer the interior gaze with conspicuous, unsettling smiles.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 격자무늬 나무 창 in the middle-center of the frame, foreground; 금월도 마을 청년들 in the middle-center of the frame, background, looks toward dining room interior.\n- KEY BACKGROUND ELEMENTS: 격자무늬 나무 창 (식당 안에서 바깥을 내다볼 수 있는 상태) — 창의 실내 쪽 면과 격자가 전경에 보이고, 그 너머 골목의 청년들이 보인다; used as The camera looks through its grid, dividing the watching faces into confined compartments; 식당 밖 골목 (청년들이 식당을 감시하고 있는 상태); used as Provides the exterior depth plane immediately behind the watching youths.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S63sh2__bgfirst_bg.png",
  "asset_id": "1ea7e225-caf2-46f6-8208-c55cb0a930d9",
  "input_asset_ids": [
   "25dcb7fb-8029-4c14-8a53-ad28de3d442e",
   "9d21e2bc-dc42-4f25-abfc-0a5069ec9b16",
   "62dfe2a8-f903-49b8-ab08-895e969c8bdf"
  ]
 },
 "S63sh2": {
  "input_fingerprint": "633d133b7f6f5533",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 창문 밖 골목에 선 금월도 마을 청년들이 식당 안쪽을 향해 기괴한 웃음을 짓는 찰나.\n\nLOCATION (lock): At the restaurant's lattice window, looking from the lit dining room toward the nighttime lane outside. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing; Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 창문 밖 골목에 선 금월도 마을 청년들이 식당 안쪽을 향해 기괴한 웃음을 짓는 찰나.\n\nLOCATION (lock): At the restaurant's lattice window, looking from the lit dining room toward the nighttime lane outside. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing; Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 창문 밖 골목에 선 금월도 마을 청년들이 식당 안쪽을 향해 기괴한 웃음을 짓는 찰나.\n\nLOCATION (lock): At the restaurant's lattice window, looking from the lit dining room toward the nighttime lane outside. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing; Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S63sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S63sh2.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L38B02.png"
    },
    {
     "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_police_patrol_boat_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 로케이션 레퍼런스(배의 갑판)를 배경으로 정확히 구현했으며, 창밖에서 안을 들여다보는 청년들의 구도와 표정도 지침을 잘 따랐습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "텍스트의 '골목' 묘사에는 충실했으나, 공간적 기준이 되는 로케이션 레퍼런스 이미지(배)를 완전히 무시하여 우선순위 3번을 위반했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 로케이션 레퍼런스(배의 갑판)를 배경으로 정확히 구현했으며, 창밖에서 안을 들여다보는 청년들의 구도와 표정도 지침을 잘 따랐습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "텍스트의 '골목' 묘사에는 충실했으나, 공간적 기준이 되는 로케이션 레퍼런스 이미지(배)를 완전히 무시하여 우선순위 3번을 위반했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "제공된 로케이션 레퍼런스(선박)를 창문 밖 배경으로 정확히 반영하고, 요구된 카메라 프레이밍과 인물들의 표정을 충실히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "첨부된 레퍼런스 이미지를 완전히 무시했으며, 화면 우측 벽면에 메뉴판 텍스트가 노출되어 '텍스트 금지' 규칙을 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "제공된 로케이션 레퍼런스(선박)를 창문 밖 배경으로 정확히 반영하고, 요구된 카메라 프레이밍과 인물들의 표정을 충실히 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "첨부된 레퍼런스 이미지를 완전히 무시했으며, 화면 우측 벽면에 메뉴판 텍스트가 노출되어 '텍스트 금지' 규칙을 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 6
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지정된 로케이션 레퍼런스(배의 갑판)를 배경으로 정확히 구현했으며, 창밖에서 안을 들여다보는 청년들의 구도와 표정도 지침을 잘 따랐습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "텍스트의 '골목' 묘사에는 충실했으나, 공간적 기준이 되는 로케이션 레퍼런스 이미지(배)를 완전히 무시하여 우선순위 3번을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L38B02.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_police_patrol_boat_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "원본 배경 이미지(SHOT BACKGROUND)에는 없는 테이블, 의자, 연필꽂이 등의 가구가 전경에 추가되었습니다. 스케치의 배경 선은 무시하고 원본 배경을 그대로 유지해야 한다는 지침을 위반했습니다.",
     "fix_en": "Remove the tables, chairs, and pencil holder from the foreground, restoring the lower framing to perfectly match the empty window sill and wall of the original background reference image.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the tables, chairs, and pencil holder from the foreground, restoring the lower framing to perfectly match the empty window sill and wall of the original background reference image.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S63sh2__bgfirst_bg.png",
   "bg_asset_id": "1ea7e225-caf2-46f6-8208-c55cb0a930d9",
   "bg_record_key": "S63sh2::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate",
   "seed_attached": true
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S63sh8": {
  "input_fingerprint": "3cdfdccf528576b0",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 붉은 피가 뚝뚝 떨어지는 생고기를 입안 한가득 쑤셔 넣는 찰나.\n\nLOCATION (lock): At the food-laden table inside the island restaurant's dining room at night. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the nighttime restaurant interior, lattice window, watching youths outside, and raw-meat table. Exclude the earlier window-centered composition and show the woman eating the meat.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains missing; fresh blood and raw meat smear her mouth and hands as she eats.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 2,
   "C": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "창문 중앙 구도를 벗어나라는 지시를 충실히 이행했으며, 펜던트 없이 생고기를 입에 쑤셔 넣는 찰나를 레퍼런스에 맞게 가장 정확히 구현했습니다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "프롬프트에 없는 타인의 손이 등장하고, 명시적으로 사라져야 할 뼈 펜던트가 여전히 목에 걸려 있어 심각한 설정 위반입니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "펜던트를 제외하고 고기를 먹는 모습은 반영했으나, 이전 숏의 창문 중앙 구도를 피하라는 지시를 어기고 그대로 중앙에 배치했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S63sh2_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "레퍼런스와 달리 창밖의 청년 5명 중 오른쪽 끝에 있던 1명이 누락되었습니다.",
     "fix_en": "Add the missing fifth young man outside the window on the far right, matching the reference.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "고기를 쥐고 있는 여성의 왼손(화면 오른쪽) 손가락이 6개입니다.",
     "fix_en": "Correct the hand on the right side of the meat to have exactly five anatomical fingers.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the missing fifth young man outside the window on the far right, matching the reference.\n- Correct the hand on the right side of the meat to have exactly five anatomical fingers.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S63sh2"
  }
 },
 "S63sh14": {
  "input_fingerprint": "e8a1c376cd671eee",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 피 묻은 손을 허공에 든 채 공포에 질려 비명을 내지르는 찰나.\n\nLOCATION (lock): Beside the raw-meat dinner table in the restaurant's lit main room. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the restaurant, meat-laden table, night lighting, and observers beyond the window. Exclude the active eating pose and show the woman recoiling from her bloodied hands.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s mouth and both hands remain covered in raw-meat blood, and her pendant remains missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "행동과 환경 연속성은 가장 잘 유지했으나, 카메라 구도를 변경하라는 참조 지시를 어기고 동일한 프레임을 복사했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "참조 이미지에 없던 벽면 조명을 두 개나 새롭게 생성하여 환경 연속성을 훼손한 심각한 위반(Hard Violation)이 있습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "왼팔에 오른손이, 오른팔에 왼손이 달린 불가능한 해부학적 오류와 존재하지 않던 스탠드 조명을 만들어낸 심각한 위반입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S63sh8_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "화면 왼쪽, 여자의 오른손 바깥쪽 가장자리에 여분의 두꺼운 살점(또는 융합된 손가락)이 있어 손의 해부학적 구조가 기형적입니다.",
     "fix_en": "Remove the extra fleshy mass on the outer left edge of the right hand (on the left side of the image) and correct the shape to be a normal human hand.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the extra fleshy mass on the outer left edge of the right hand (on the left side of the image) and correct the shape to be a normal human hand.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S63sh8"
  }
 },
 "S64sh2::bgfirst_bg": {
  "input_fingerprint": "7c377467a3556791",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 깜깜한 해송 숲길에서 한쪽 발을 높이 든 채 앞으로 힘차게 내딛는 자세의 강영란(흰 머리 여자) 전신.\n\nLOCATION (lock): On a pitch-dark pine-forest path leading uphill from the village beneath the nearly full moon.\n\nTIME OF DAY (lock): night, near-full moon and starless sky.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track parallel from beside the uphill path at waist height, framing 강영란(흰 머리 여자) head to toe in the middle of a wide image as her lifted foot drives forward. Match her pace for this single stride so her body stays stable while the path and escorting 금월도 마을 주민 slide laterally behind and around her.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the middle-center of the frame, midground, moves toward uphill forest path; 호위하는 금월도 마을 주민 in the middle-left of the frame, midground, moves toward uphill forest path; uphill forest path in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 해송 숲길 (일행이 올라가는 길); used as Creates the lateral movement axis and visibly rises in the direction of travel; 해송 숲 (불빛 없이 어두운 상태); used as Forms the enclosing background around the moving group; 보름달 (거의 완성된 상태); used as Held above the tree line when composition permits, linking this movement to the completed lunar approach.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 깜깜한 해송 숲길에서 한쪽 발을 높이 든 채 앞으로 힘차게 내딛는 자세의 강영란(흰 머리 여자) 전신.\n\nLOCATION (lock): On a pitch-dark pine-forest path leading uphill from the village beneath the nearly full moon.\n\nTIME OF DAY (lock): night, near-full moon and starless sky.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track parallel from beside the uphill path at waist height, framing 강영란(흰 머리 여자) head to toe in the middle of a wide image as her lifted foot drives forward. Match her pace for this single stride so her body stays stable while the path and escorting 금월도 마을 주민 slide laterally behind and around her.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the middle-center of the frame, midground, moves toward uphill forest path; 호위하는 금월도 마을 주민 in the middle-left of the frame, midground, moves toward uphill forest path; uphill forest path in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 해송 숲길 (일행이 올라가는 길); used as Creates the lateral movement axis and visibly rises in the direction of travel; 해송 숲 (불빛 없이 어두운 상태); used as Forms the enclosing background around the moving group; 보름달 (거의 완성된 상태); used as Held above the tree line when composition permits, linking this movement to the completed lunar approach.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S64sh2__bgfirst_bg.png",
  "asset_id": "3f3c9497-c1c4-4d65-a238-5ad6a42f972e",
  "input_asset_ids": [
   "c5c445c2-5db8-4006-a13a-1eebba0ae06a",
   "5b5030a4-55f4-4366-af70-98b1d535b160"
  ]
 },
 "S64sh2": {
  "input_fingerprint": "2a4ba6c9dbaf1ed3",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, near-full moon and starless sky.\n\nSHOT TEXT (authoritative, Korean): 깜깜한 해송 숲길에서 한쪽 발을 높이 든 채 앞으로 힘차게 내딛는 자세의 강영란(흰 머리 여자) 전신.\n\nLOCATION (lock): On a pitch-dark pine-forest path leading uphill from the village beneath the nearly full moon. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran continues to wear the red shaman’s robe prepared at the restaurant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, near-full moon and starless sky.\n\nSHOT TEXT (authoritative, Korean): 깜깜한 해송 숲길에서 한쪽 발을 높이 든 채 앞으로 힘차게 내딛는 자세의 강영란(흰 머리 여자) 전신.\n\nLOCATION (lock): On a pitch-dark pine-forest path leading uphill from the village beneath the nearly full moon. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran continues to wear the red shaman’s robe prepared at the restaurant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, near-full moon and starless sky.\n\nSHOT TEXT (authoritative, Korean): 깜깜한 해송 숲길에서 한쪽 발을 높이 든 채 앞으로 힘차게 내딛는 자세의 강영란(흰 머리 여자) 전신.\n\nLOCATION (lock): On a pitch-dark pine-forest path leading uphill from the village beneath the nearly full moon. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran continues to wear the red shaman’s robe prepared at the restaurant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S64sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S64sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1010229>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L18B03.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1010229>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "레퍼런스의 장소 디테일(붉은 천, 조명)과 무녀복의 금박 무늬를 매우 정확히 재현하고 포즈를 잘 살렸으나, 하늘에 별이 일부 묘사된 점은 아쉽습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "요청된 포즈와 달은 묘사되었으나, 지정된 장소의 식별 요소가 누락되었고 무녀복의 디테일 없이 단순한 붉은 한복으로 표현되어 일치도가 낮습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "레퍼런스의 장소 디테일(붉은 천, 조명)과 무녀복의 금박 무늬를 매우 정확히 재현하고 포즈를 잘 살렸으나, 하늘에 별이 일부 묘사된 점은 아쉽습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "요청된 포즈와 달은 묘사되었으나, 지정된 장소의 식별 요소가 누락되었고 무녀복의 디테일 없이 단순한 붉은 한복으로 표현되어 일치도가 낮습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "장소 레퍼런스의 특징(등불, 밧줄)과 의상 디테일을 잘 살렸으며, 지정된 한 발을 높이 든 역동적인 전신 포즈를 충실하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "지정된 숲길의 주요 설치물(등불, 밧줄)이 모두 누락되었으며, 레퍼런스의 화려한 무당복이 단순한 붉은 한복으로 변형되어 로케이션 및 의상 조건을 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "장소 레퍼런스의 특징(등불, 밧줄)과 의상 디테일을 잘 살렸으며, 지정된 한 발을 높이 든 역동적인 전신 포즈를 충실하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지정된 숲길의 주요 설치물(등불, 밧줄)이 모두 누락되었으며, 레퍼런스의 화려한 무당복이 단순한 붉은 한복으로 변형되어 로케이션 및 의상 조건을 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 7,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 7,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "레퍼런스의 장소 디테일(붉은 천, 조명)과 무녀복의 금박 무늬를 매우 정확히 재현하고 포즈를 잘 살렸으나, 하늘에 별이 일부 묘사된 점은 아쉽습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "요청된 포즈와 달은 묘사되었으나, 지정된 장소의 식별 요소가 누락되었고 무녀복의 디테일 없이 단순한 붉은 한복으로 표현되어 일치도가 낮습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L18B03.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1010229>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "우측 상단 하늘에 별이 보입니다 (별이 없는 하늘 지시 위반).",
     "fix_en": "Remove the stars from the sky in the top right corner.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "레퍼런스에 없는 붉은 두건을 쓰고 있으며, 머리에 써야 할 깃털 모자를 왼손에 들고 있습니다.",
     "fix_en": "Remove the red bandana from her head and the hat from her left hand, and instead place the peacock feather hat properly on her head.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "오른손에 프롬프트에 명시되지 않은 금속 방울을 들고 있습니다.",
     "fix_en": "Remove the golden bell from her right hand so that her hand is empty.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "신발이 레퍼런스의 붉은 전통 신발과 달리 검은색입니다.",
     "fix_en": "Change the black shoes to match the red traditional shoes from the character reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the stars from the sky in the top right corner.\n- Remove the red bandana from her head and the hat from her left hand, and instead place the peacock feather hat properly on her head.\n- Remove the golden bell from her right hand so that her hand is empty.\n- Change the black shoes to match the red traditional shoes from the character reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S64sh2__bgfirst_bg.png",
   "bg_asset_id": "3f3c9497-c1c4-4d65-a238-5ad6a42f972e",
   "bg_record_key": "S64sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S65sh2::bgfirst_bg": {
  "input_fingerprint": "3bbb4976064d2b63",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 화재 경보 소리에 화들짝 놀라 식당 뒤편으로 고개를 홱 돌린 금월도 마을 청년들의 찰나.\n\nLOCATION (lock): On the nighttime village lane directly outside the island restaurant.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the youths' front-side at eye level, whip the pan with their abrupt head turn and settle on a shared medium frame of their upper bodies. Their torsos still angle toward the restaurant while every head snaps toward the rear direction at frame right, capturing the split second before they break into a run.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 금월도 마을 청년들 in the middle-center of the frame, midground, looks toward restaurant rear where the alarm originated; direction of the restaurant rear in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 반월 식당 (청년들이 감시하던 건물) — 식당의 골목 쪽 외벽이 카메라에 비스듬히 보이고 뒤편 방향은 화면 오른쪽으로 이어진다; used as Its side remains behind the youths, showing that their bodies have not yet followed their turned heads; 식당 밖 골목 (주민과 청년들이 요소요소 지키고 있는 상태); used as Provides the open route into which the next tracking movement will continue.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 화재 경보 소리에 화들짝 놀라 식당 뒤편으로 고개를 홱 돌린 금월도 마을 청년들의 찰나.\n\nLOCATION (lock): On the nighttime village lane directly outside the island restaurant.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the youths' front-side at eye level, whip the pan with their abrupt head turn and settle on a shared medium frame of their upper bodies. Their torsos still angle toward the restaurant while every head snaps toward the rear direction at frame right, capturing the split second before they break into a run.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 금월도 마을 청년들 in the middle-center of the frame, midground, looks toward restaurant rear where the alarm originated; direction of the restaurant rear in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 반월 식당 (청년들이 감시하던 건물) — 식당의 골목 쪽 외벽이 카메라에 비스듬히 보이고 뒤편 방향은 화면 오른쪽으로 이어진다; used as Its side remains behind the youths, showing that their bodies have not yet followed their turned heads; 식당 밖 골목 (주민과 청년들이 요소요소 지키고 있는 상태); used as Provides the open route into which the next tracking movement will continue.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S65sh2__bgfirst_bg.png",
  "asset_id": "6ce606c2-dfff-46d3-a064-d1e44465a08a",
  "input_asset_ids": [
   "2934fc5d-6b9c-4861-a318-256bc14beea7",
   "803cc956-9eff-49c4-b977-3e62ad55fc55"
  ]
 },
 "S65sh2": {
  "input_fingerprint": "037263938df0657e",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 화재 경보 소리에 화들짝 놀라 식당 뒤편으로 고개를 홱 돌린 금월도 마을 청년들의 찰나.\n\nLOCATION (lock): On the nighttime village lane directly outside the island restaurant. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The villagers’ surveillance posts around the restaurant remain in place until the fire alarm draws them away.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 화재 경보 소리에 화들짝 놀라 식당 뒤편으로 고개를 홱 돌린 금월도 마을 청년들의 찰나.\n\nLOCATION (lock): On the nighttime village lane directly outside the island restaurant. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The villagers’ surveillance posts around the restaurant remain in place until the fire alarm draws them away.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 화재 경보 소리에 화들짝 놀라 식당 뒤편으로 고개를 홱 돌린 금월도 마을 청년들의 찰나.\n\nLOCATION (lock): On the nighttime village lane directly outside the island restaurant. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The villagers’ surveillance posts around the restaurant remain in place until the fire alarm draws them away.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S65sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S65sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:902182>"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:902182>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "레퍼런스에 제시된 특정 인물(복장 포함)과 장소의 표지석들을 정확하게 포함시켰으며, 경보 소리에 놀라 뒤를 돌아보는 인물들의 행동 지침을 충실히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지정된 인물 레퍼런스의 외모와 복장이 누락되었으며, 장소 레퍼런스의 핵심인 표지석 구조물도 전혀 반영되지 않아 주요 지침을 위반했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "레퍼런스에 제시된 특정 인물(복장 포함)과 장소의 표지석들을 정확하게 포함시켰으며, 경보 소리에 놀라 뒤를 돌아보는 인물들의 행동 지침을 충실히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지정된 인물 레퍼런스의 외모와 복장이 누락되었으며, 장소 레퍼런스의 핵심인 표지석 구조물도 전혀 반영되지 않아 주요 지침을 위반했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 캐릭터의 의상과 외모를 매우 정확하게 구현했으며, 로케이션 구조물의 특징과 인물들의 행동 지시를 잘 반영했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "캐릭터 레퍼런스의 의상과 외모가 전혀 반영되지 않았으며, 로케이션 사진의 핵심 구조물(비석)도 누락되었습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 캐릭터의 의상과 외모를 매우 정확하게 구현했으며, 로케이션 구조물의 특징과 인물들의 행동 지시를 잘 반영했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "캐릭터 레퍼런스의 의상과 외모가 전혀 반영되지 않았으며, 로케이션 사진의 핵심 구조물(비석)도 누락되었습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "레퍼런스에 제시된 특정 인물(복장 포함)과 장소의 표지석들을 정확하게 포함시켰으며, 경보 소리에 놀라 뒤를 돌아보는 인물들의 행동 지침을 충실히 구현했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "지정된 인물 레퍼런스의 외모와 복장이 누락되었으며, 장소 레퍼런스의 핵심인 표지석 구조물도 전혀 반영되지 않아 주요 지침을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:902182>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "위치 레퍼런스의 배경은 흙길과 소나무 숲이지만, 포장도로와 상업 건물로 완전히 다르게 생성되어 영구적인 현장 세부 정보(permanent site details)가 무시되었습니다.",
     "fix_en": "Replace the paved road and buildings with a dirt path surrounded by a dark pine forest, matching the exact permanent site details of the location reference.",
     "needs_regeneration": true,
     "unfixable": true
    },
    {
     "issue_ko": "식당의 메인 간판, 세로 간판, 그리고 창문에 의미를 알 수 없거나 철자가 틀린 텍스트가 생성되었습니다.",
     "fix_en": "Remove all text from the restaurant's top sign, vertical side sign, and windows.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "왼쪽 작은 비석의 '수리영' 글자 아래에 레퍼런스에는 없는 잘못된 텍스트가 추가되었습니다.",
     "fix_en": "Remove the garbled text inside the parentheses below '수리영' on the left small stone marker.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "오른쪽 큰 비석 받침대의 텍스트가 레퍼런스와 다르게 뭉개져서 알아볼 수 없습니다.",
     "fix_en": "Replace the garbled text on the base of the right large stone marker to clearly read '수리영 2012.10.1'.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all text from the restaurant's top sign, vertical side sign, and windows.\n- Remove the garbled text inside the parentheses below '수리영' on the left small stone marker.\n- Replace the garbled text on the base of the right large stone marker to clearly read '수리영 2012.10.1'.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S65sh2__bgfirst_bg.png",
   "bg_asset_id": "6ce606c2-dfff-46d3-a064-d1e44465a08a",
   "bg_record_key": "S65sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S66sh1": {
  "input_fingerprint": "713ae2cf7642ad89",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, moon partly obscured by smoke.\n\nSHOT TEXT (authoritative, Korean): 낡은 창고 위로 거대한 불길이 솟구친 채 붉게 타오르는 뒤뜰의 전경.\n\nLOCATION (lock): In the restaurant's rear yard, where the old storage shed is engulfed in flames and black smoke. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The backyard shed is actively burning, producing dense black smoke.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 9,
   "B": 5,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 9,
    "verdict_ko": "지정된 구조물(오른쪽)의 형태와 재질을 완벽하게 반영했으며, 프롬프트대로 왼쪽의 낡은 창고에만 불길을 묘사하여 공간적 일관성을 유지했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "지정된 구조물로 교체하는 데는 성공했으나, 타올라야 할 왼쪽 창고가 아닌 본건물(오른쪽)에 불을 붙여 샷의 핵심 상황을 왜곡했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "오른쪽 구조물을 레퍼런스 이미지로 교체하지 않고 로케이션 사진의 원래 건물(염소 우리 포함)을 그대로 복사하여 구조물 지시를 완전히 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L53B02.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_restaurant_property_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "로케이션 사진의 기존 콘크리트 건물과 가축 우리가 완전히 제거되지 않고 새 구조물(지정된 레퍼런스)의 왼쪽에 어색하게 덧붙여져 있어, 건물의 전체적인 형태와 구조가 잘못 생성되었습니다.",
     "fix_en": "Remove the remnant concrete wall and animal pen in the middle, and extend the new wood-and-stone building to properly replace the entire original structure.",
     "needs_regeneration": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the remnant concrete wall and animal pen in the middle, and extend the new wood-and-stone building to properly replace the entire original structure.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트+seed만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_bypass:bg_only"
 },
 "S67sh3::bgfirst_bg": {
  "input_fingerprint": "5a6d57ecde7aa7b1",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 갑작스러운 플래시 불빛에 눈이 부셔 미간을 찡그리며 화들짝 놀란 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): On a dark uphill forest path, illuminated by a newly switched-on flashlight and moonlight.\n\nTIME OF DAY (lock): night, moonlight.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the frontal dolly-in directly before 김지후(강수리영) at her eye level, isolating her face in a tight close frame as the sudden flashlight beam makes her flinch and squeeze her eyes nearly shut. Her shoulders remain angled by the uphill motion and a narrow strip of the path survives behind her, preventing the reaction from reading as a detached portrait.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 김지후(강수리영) in the middle-center of the frame, foreground, looks toward 정인우 holding the flashlight near the camera axis.\n- KEY BACKGROUND ELEMENTS: 숲길 (김지후(강수리영)와 정인우가 올라가는 상태); used as Retained as a narrow, soft background strip to preserve her uphill movement; 플래시 라이트 불빛 (갑자기 켜져 얼굴에 쏟아지는 상태) — 정인우 쪽에서 김지후(강수리영)의 얼굴을 향해 정면에 가깝게 비춘다; used as The beam strikes her face from near the camera axis and motivates the squinting reaction.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 갑작스러운 플래시 불빛에 눈이 부셔 미간을 찡그리며 화들짝 놀란 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): On a dark uphill forest path, illuminated by a newly switched-on flashlight and moonlight.\n\nTIME OF DAY (lock): night, moonlight.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the frontal dolly-in directly before 김지후(강수리영) at her eye level, isolating her face in a tight close frame as the sudden flashlight beam makes her flinch and squeeze her eyes nearly shut. Her shoulders remain angled by the uphill motion and a narrow strip of the path survives behind her, preventing the reaction from reading as a detached portrait.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 김지후(강수리영) in the middle-center of the frame, foreground, looks toward 정인우 holding the flashlight near the camera axis.\n- KEY BACKGROUND ELEMENTS: 숲길 (김지후(강수리영)와 정인우가 올라가는 상태); used as Retained as a narrow, soft background strip to preserve her uphill movement; 플래시 라이트 불빛 (갑자기 켜져 얼굴에 쏟아지는 상태) — 정인우 쪽에서 김지후(강수리영)의 얼굴을 향해 정면에 가깝게 비춘다; used as The beam strikes her face from near the camera axis and motivates the squinting reaction.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S67sh3__bgfirst_bg.png",
  "asset_id": "27c9efe1-2ded-4d15-ad5a-1ac0ae80b705",
  "input_asset_ids": [
   "bb018548-412f-43df-84f9-bb859e18999b",
   "947f1562-e8ef-43fc-ad1a-3ca896912adb"
  ]
 },
 "S67sh3": {
  "input_fingerprint": "5eb183f199e2a670",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, moonlight.\n\nSHOT TEXT (authoritative, Korean): 갑작스러운 플래시 불빛에 눈이 부셔 미간을 찡그리며 화들짝 놀란 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): On a dark uphill forest path, illuminated by a newly switched-on flashlight and moonlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Residual blood from Jihu’s raw-meat binge remains on her mouth and hands; her pendant is still missing, and Inwoo’s arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, moonlight.\n\nSHOT TEXT (authoritative, Korean): 갑작스러운 플래시 불빛에 눈이 부셔 미간을 찡그리며 화들짝 놀란 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): On a dark uphill forest path, illuminated by a newly switched-on flashlight and moonlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Residual blood from Jihu’s raw-meat binge remains on her mouth and hands; her pendant is still missing, and Inwoo’s arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, moonlight.\n\nSHOT TEXT (authoritative, Korean): 갑작스러운 플래시 불빛에 눈이 부셔 미간을 찡그리며 화들짝 놀란 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): On a dark uphill forest path, illuminated by a newly switched-on flashlight and moonlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Residual blood from Jihu’s raw-meat binge remains on her mouth and hands; her pendant is still missing, and Inwoo’s arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S67sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S67sh3.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L23B06.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "플래시 불빛에 눈이 부셔 미간을 찡그리는 표정과 빛을 가리는 자연스러운 손동작이 클로즈업 샷으로 훌륭하게 연출되었습니다."
     },
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "놀란 표정과 입가, 손의 핏자국 등 주요 설정은 잘 반영되었으나, 손의 포즈가 상대적으로 다소 부자연스럽게 보입니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "플래시 불빛에 눈이 부셔 미간을 찡그리는 표정과 빛을 가리는 자연스러운 손동작이 클로즈업 샷으로 훌륭하게 연출되었습니다."
     },
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "놀란 표정과 입가, 손의 핏자국 등 주요 설정은 잘 반영되었으나, 손의 포즈가 상대적으로 다소 부자연스럽게 보입니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "플래시 불빛에 눈이 부셔 놀란 표정과 빛을 가리려는 손동작이 지시문과 정확히 일치하며, 피 묻은 분장과 배경도 충실하게 구현되었습니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "눈부심을 표현했으나 양손의 위치와 포즈가 화들짝 놀란 반응이라기보다 다소 어색하며, 손가락 형태가 자연스럽지 않습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "플래시 불빛에 눈이 부셔 놀란 표정과 빛을 가리려는 손동작이 지시문과 정확히 일치하며, 피 묻은 분장과 배경도 충실하게 구현되었습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "눈부심을 표현했으나 양손의 위치와 포즈가 화들짝 놀란 반응이라기보다 다소 어색하며, 손가락 형태가 자연스럽지 않습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 13,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 13,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "플래시 불빛에 눈이 부셔 미간을 찡그리는 표정과 빛을 가리는 자연스러운 손동작이 클로즈업 샷으로 훌륭하게 연출되었습니다."
   },
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "놀란 표정과 입가, 손의 핏자국 등 주요 설정은 잘 반영되었으나, 손의 포즈가 상대적으로 다소 부자연스럽게 보입니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L23B06.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "올려진 손의 손가락이 6개로 묘사되어 해부학적으로 잘못되었습니다.",
     "fix_en": "Remove the extra finger between the ring finger and the pinky on the raised hand so it has exactly five normal digits.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the extra finger between the ring finger and the pinky on the raised hand so it has exactly five normal digits.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S67sh3__bgfirst_bg.png",
   "bg_asset_id": "27c9efe1-2ded-4d15-ad5a-1ac0ae80b705",
   "bg_record_key": "S67sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S67sh8": {
  "input_fingerprint": "e63c45967d479690",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, moonlight.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 텅 빈 자신의 목덜미에 손을 얹은 채 불안한 눈빛을 띠고 있는 찰나.\n\nLOCATION (lock): Along the nighttime pine-forest trail as the pair move away from the village. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the dark pine path, handheld flashlight, moonlight, and the pair's continuing clothes. Exclude the sudden face-lit reaction and show the woman checking her empty neck.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s neck remains bare because her antler-bone pendant is missing.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 8,
   "A": 4,
   "B": 2
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "목덜미를 확인하는 불안한 표정을 가장 정확히 연출했으며, 금지된 추가 인물 없이 상황을 훌륭하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "화면 우측에 프롬프트에 명시되지 않은 타인의 손이 등장하여 '지시되지 않은 인물 등장 불가' 규칙을 위반했습니다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "인물 뒤편에서 튀어나오는 기형적이고 해부학적으로 불가능한 타인의 팔이 묘사되어 심각한 위반입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S67sh3_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "배경의 오솔길 바닥에 떨어져 빛나고 있는 광원이 잘못 생성되었습니다 (손전등은 카메라 밖에서 인물을 비추고 있어야 합니다).",
     "fix_en": "Remove the glowing white object from the path in the background.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the glowing white object from the path in the background.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S67sh3"
  }
 },
 "S67sh9": {
  "input_fingerprint": "86e75c23305fa0cf",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, moonlight.\n\nSHOT TEXT (authoritative, Korean): 어두운 숲길에서 심각한 표정으로 우뚝 멈춰 선 정인우와 김지후(강수리영)가 굳은 얼굴로 서로를 마주 보는 전경.\n\nLOCATION (lock): On the unlit forest path where the pair pause beneath the trees. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the narrow forest path, flashlight beam, darkness, and the two walkers' positions. Exclude the neck-checking gesture and show them stopped face-to-face.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains missing, and Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "지시된 전경(Full shot) 프레이밍을 정확히 구현했으며, 남자의 깁스 형태와 손전등 조명, 인물의 위치 및 복장이 참조 이미지와 자연스럽게 일치합니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "두 사람이 마주보는 구도와 배경은 적절하나, 남자의 깁스가 명확하지 않고 손전등의 빛이 닿는 곳에 조명 효과가 생략되어 물리적으로 어색합니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "남자의 손에 든 손전등 외에 허리춤에서 또 다른 빛이 뿜어져 나오는 형태의 치명적인 렌더링 오류(Hard Violation)가 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S67sh8_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "남자의 복부에서 손전등을 쥔 세 번째 손이 튀어나와 있습니다.",
     "fix_en": "Remove the extra hand and the flashlight protruding from the man's torso.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "레퍼런스 이미지에서 여자의 머리는 풀려 있으나, 이 이미지에서는 포니테일로 묶여 있습니다.",
     "fix_en": "Change the woman's hair to be loose and untied, removing the ponytail to match the reference.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "레퍼런스 이미지에 있는 여자의 아랫입술 부근 피/상처 자국이 누락되었습니다.",
     "fix_en": "Add the blood or dirt smudge to the woman's lower lip to match the reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the extra hand and the flashlight protruding from the man's torso.\n- Change the woman's hair to be loose and untied, removing the ponytail to match the reference.\n- Add the blood or dirt smudge to the woman's lower lip to match the reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S67sh8"
  }
 },
 "S68sh1::bgfirst_bg": {
  "input_fingerprint": "0b050cdd9d5c4f7b",
  "prompt": "Turn the attached storyboard sketch into a photorealistic film still of\nits LOCATION ONLY, keeping the figures exactly as they are.\n\nKEEP EXACTLY: the camera framing, the horizon, and where every element\nsits in the frame. Each mannequin figure stays a plain grey featureless\nmannequin standing in the very same spot, at the same size, in the same\npose, turned the same way — do not turn them into people, do not move,\nrotate, mirror or re-pose them, do not add or remove figures.\n\nBUILD PHOTOREALISTICALLY: everything that is not a figure — ground,\nsurfacing, structures, vegetation, sky, water, distance. The bare lines\nof the sketch are a layout guide; replace them with the real materials,\ndepth and lighting of the place described below.\n\nSHOT TEXT this background must serve (Korean): 모닥불이 둥글게 타오르는 숲속 제단의 거대한 바위 앞에 강영란(흰 머리 여자)이 무릎을 꿇고 앉은 뒷모습.\n\nLOCATION (lock): In the ritual clearing before the huge boulder, encircled by burning campfires and drifting smoke.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Begin at standing eye level several meters behind 강영란(흰 머리 여자), centered on the axis connecting her kneeling back to the enormous rock, and start a restrained dolly forward from outside the ring of fires. Keep her full kneeling figure low in the central frame while the rock and circular ritual layout remain clearly readable beyond her.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the lower-center of the frame, midground, looks toward large rock and altar; large rock and altar in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 거대한 바위 (제단 뒤에 놓여 있음); used as Central ritual destination aligned directly beyond the kneeling figure; 여러 개의 모닥불 (둥근 원을 이루며 활활 타오름); used as Establishes the circular boundary around the ritual space; 연기 (희뿌연 안개처럼 주변을 감쌈); used as Occupies the space around the fires and rock, softening their outlines; 제단 (바위 앞에 놓여 있음) — Its front edge faces the camera beyond 강영란(흰 머리 여자); used as Marks the immediate ritual area between the kneeling figure and rock.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nTHINGS AT THIS PLACE:\nThe site is organized along a forest path that enters from the outer forest zone and proceeds inward through a dense stand of pine trunks. At its inner end, the recognizable path disappears into a trackless deep-forest interior. Beyond and around this transition, several stone markers inscribed with Suri-young's name are distributed at separate points along the continuing walking line. Dense pine trunks enclose the route on both sides and continue past the path disappearance point. Undergrowth fills the spaces between the trunks, borders the path, and occupies the off-path ground around the dispersed marker stones.\n- path: Outer-to-deep forest path\n- path end: Forest path disappearance point\n- marker stone: Group of stone markers inscribed with Suri-young's name\n- tree stand: Dense pine trunk stand\n- undergrowth: Intervening forest undergrowth\n\nWORLD FACTS (creator-confirmed — always true):\n- Region (real-world reference): 대한민국 서울권과 서남해안 일대, 대한민국 영토 내 서해의 가상 외딴섬 공동체\n- Era: 주 배경은 스마트폰과 현대 경찰·교통 체계가 존재하는 2020년대 중반의 현대이며, 1670년 조선시대와 20세기 전반 일제강점기 전후의 역사적 재현 구간이 명확히 구분되어 병존한다.\n- 현대 구간은 2020년대 대한민국의 생활상으로, 1670년대 구간은 조선 후기의 복식과 생활 도구로, 20세기 전반 구간은 당시 조선인 복식과 일본군 장비로 표현한다. 시대 전환이 없는 프레임에는 서로 다른 시대의 의상·무기·교통수단·전자기기를 혼합하지 않는다.\n- 현대 육지에서는 스마트폰, 메신저 앱, 지도 앱, CCTV, 바코드 출근 시스템, 현대 경찰 장비, 승용차·구형 SUV·동력 어선이 자연스럽게 존재한다. 섬에서는 휴대전화 신호와 인터넷 연결이 불안정하거나 차단되며, 제한적인 전기 조명과 오래된 생활 설비를 우선하되 현대 물품 자체가 전혀 없는 원시사회로 오해하지 않는다. 역사 구간에는 디지털 기기와 현대 차량을 배제한다.\n\nTHINGS THAT LIVE AT THIS PLACE (what they are, not where they go): the\nlist above names what permanently belongs to this site. Where each one\nsits in this frame is decided by the attached sketch alone — do not add\nor reposition anything on the strength of the list, build only what the\nsketch already has lines for, and do not invent facilities that are not\nlisted.\n\nNo text, no captions, no diagram symbols, no markers anywhere in the\nimage.",
  "effective_prompt": "Turn the attached storyboard sketch into a photorealistic film still of\nits LOCATION ONLY, keeping the figures exactly as they are.\n\nKEEP EXACTLY: the camera framing, the horizon, and where every element\nsits in the frame. Each mannequin figure stays a plain grey featureless\nmannequin standing in the very same spot, at the same size, in the same\npose, turned the same way — do not turn them into people, do not move,\nrotate, mirror or re-pose them, do not add or remove figures.\n\nBUILD PHOTOREALISTICALLY: everything that is not a figure — ground,\nsurfacing, structures, vegetation, sky, water, distance. The bare lines\nof the sketch are a layout guide; replace them with the real materials,\ndepth and lighting of the place described below.\n\nSHOT TEXT this background must serve (Korean): 모닥불이 둥글게 타오르는 숲속 제단의 거대한 바위 앞에 강영란(흰 머리 여자)이 무릎을 꿇고 앉은 뒷모습.\n\nLOCATION (lock): In the ritual clearing before the huge boulder, encircled by burning campfires and drifting smoke.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Begin at standing eye level several meters behind 강영란(흰 머리 여자), centered on the axis connecting her kneeling back to the enormous rock, and start a restrained dolly forward from outside the ring of fires. Keep her full kneeling figure low in the central frame while the rock and circular ritual layout remain clearly readable beyond her.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the lower-center of the frame, midground, looks toward large rock and altar; large rock and altar in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 거대한 바위 (제단 뒤에 놓여 있음); used as Central ritual destination aligned directly beyond the kneeling figure; 여러 개의 모닥불 (둥근 원을 이루며 활활 타오름); used as Establishes the circular boundary around the ritual space; 연기 (희뿌연 안개처럼 주변을 감쌈); used as Occupies the space around the fires and rock, softening their outlines; 제단 (바위 앞에 놓여 있음) — Its front edge faces the camera beyond 강영란(흰 머리 여자); used as Marks the immediate ritual area between the kneeling figure and rock.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nTHINGS AT THIS PLACE:\nThe site is organized along a forest path that enters from the outer forest zone and proceeds inward through a dense stand of pine trunks. At its inner end, the recognizable path disappears into a trackless deep-forest interior. Beyond and around this transition, several stone markers inscribed with Suri-young's name are distributed at separate points along the continuing walking line. Dense pine trunks enclose the route on both sides and continue past the path disappearance point. Undergrowth fills the spaces between the trunks, borders the path, and occupies the off-path ground around the dispersed marker stones.\n- path: Outer-to-deep forest path\n- path end: Forest path disappearance point\n- marker stone: Group of stone markers inscribed with Suri-young's name\n- tree stand: Dense pine trunk stand\n- undergrowth: Intervening forest undergrowth\n\nWORLD FACTS (creator-confirmed — always true):\n- Region (real-world reference): 대한민국 서울권과 서남해안 일대, 대한민국 영토 내 서해의 가상 외딴섬 공동체\n- Era: 주 배경은 스마트폰과 현대 경찰·교통 체계가 존재하는 2020년대 중반의 현대이며, 1670년 조선시대와 20세기 전반 일제강점기 전후의 역사적 재현 구간이 명확히 구분되어 병존한다.\n- 현대 구간은 2020년대 대한민국의 생활상으로, 1670년대 구간은 조선 후기의 복식과 생활 도구로, 20세기 전반 구간은 당시 조선인 복식과 일본군 장비로 표현한다. 시대 전환이 없는 프레임에는 서로 다른 시대의 의상·무기·교통수단·전자기기를 혼합하지 않는다.\n- 현대 육지에서는 스마트폰, 메신저 앱, 지도 앱, CCTV, 바코드 출근 시스템, 현대 경찰 장비, 승용차·구형 SUV·동력 어선이 자연스럽게 존재한다. 섬에서는 휴대전화 신호와 인터넷 연결이 불안정하거나 차단되며, 제한적인 전기 조명과 오래된 생활 설비를 우선하되 현대 물품 자체가 전혀 없는 원시사회로 오해하지 않는다. 역사 구간에는 디지털 기기와 현대 차량을 배제한다.\n\nTHINGS THAT LIVE AT THIS PLACE (what they are, not where they go): the\nlist above names what permanently belongs to this site. Where each one\nsits in this frame is decided by the attached sketch alone — do not add\nor reposition anything on the strength of the list, build only what the\nsketch already has lines for, and do not invent facilities that are not\nlisted.\n\nNo text, no captions, no diagram symbols, no markers anywhere in the\nimage.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S68sh1__bgfirst_bg.png",
  "asset_id": "bcdee313-035b-4bbe-8cd6-0ac312f2c171",
  "input_asset_ids": [
   "a2d138d1-57cd-468c-bc5e-47398e100879"
  ]
 },
 "S68sh1": {
  "input_fingerprint": "ac1f7bd3da7c5e32",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 모닥불이 둥글게 타오르는 숲속 제단의 거대한 바위 앞에 강영란(흰 머리 여자)이 무릎을 꿇고 앉은 뒷모습.\n\nLOCATION (lock): In the ritual clearing before the huge boulder, encircled by burning campfires and drifting smoke. The shot takes place here — the attached STORYBOARD SKETCH fixes the staging, camera and figure placement of this exact place. No location photograph is attached — build the location itself strictly from the location text above and the shot text, inventing nothing beyond them.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The giant ritual rock remains stained with dark-red animal blood, and Youngran remains dressed in her red shaman’s robe.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Replace every grey or outlined mannequin figure in the FIRST attached\nimage with the real people the CHARACTER REFERENCE images show, and\noutput a photorealistic film still. Each mannequin becomes the character\nwhose POSE CANON and the shot text match that figure's pose and\nposition; never repeat one person across two mannequins. If no CHARACTER\nREFERENCE image is attached for a figure, still replace it with a\nplausible real person consistent with the text. Every mannequin becomes\na person — erasing a figure, or leaving one out, is not a replacement,\nand the number of figures never changes.\n\nKEEP EXACTLY: the background of the first image, the camera framing,\nand each mannequin's position, size, pose and the direction it is\nturned. A person must be exactly where their mannequin was, at the same\nscale, in the same pose, facing the same way — never mirrored, never\nre-staged, never straightened up, stood upright or otherwise re-posed.\nNo mannequin, grey figure or sketch line may remain anywhere in the\noutput.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 모닥불이 둥글게 타오르는 숲속 제단의 거대한 바위 앞에 강영란(흰 머리 여자)이 무릎을 꿇고 앉은 뒷모습.\n\nLOCATION (lock): In the ritual clearing before the huge boulder, encircled by burning campfires and drifting smoke. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The giant ritual rock remains stained with dark-red animal blood, and Youngran remains dressed in her red shaman’s robe.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Replace every grey or outlined mannequin figure in the FIRST attached\nimage with the real people the CHARACTER REFERENCE images show, and\noutput a photorealistic film still. Each mannequin becomes the character\nwhose POSE CANON and the shot text match that figure's pose and\nposition; never repeat one person across two mannequins. If no CHARACTER\nREFERENCE image is attached for a figure, still replace it with a\nplausible real person consistent with the text. Every mannequin becomes\na person — erasing a figure, or leaving one out, is not a replacement,\nand the number of figures never changes.\n\nKEEP EXACTLY: the background of the first image, the camera framing,\nand each mannequin's position, size, pose and the direction it is\nturned. A person must be exactly where their mannequin was, at the same\nscale, in the same pose, facing the same way — never mirrored, never\nre-staged, never straightened up, stood upright or otherwise re-posed.\nNo mannequin, grey figure or sketch line may remain anywhere in the\noutput.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 모닥불이 둥글게 타오르는 숲속 제단의 거대한 바위 앞에 강영란(흰 머리 여자)이 무릎을 꿇고 앉은 뒷모습.\n\nLOCATION (lock): In the ritual clearing before the huge boulder, encircled by burning campfires and drifting smoke. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The giant ritual rock remains stained with dark-red animal blood, and Youngran remains dressed in her red shaman’s robe.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "C": "Replace every grey or outlined mannequin figure in the FIRST attached\nimage with the real people the CHARACTER REFERENCE images show, and\noutput a photorealistic film still. Each mannequin becomes the character\nwhose POSE CANON and the shot text match that figure's pose and\nposition; never repeat one person across two mannequins. If no CHARACTER\nREFERENCE image is attached for a figure, still replace it with a\nplausible real person consistent with the text. Every mannequin becomes\na person — erasing a figure, or leaving one out, is not a replacement,\nand the number of figures never changes.\n\nKEEP EXACTLY: the background of the first image, the camera framing,\nand each mannequin's position, size, pose and the direction it is\nturned. A person must be exactly where their mannequin was, at the same\nscale, in the same pose, facing the same way — never mirrored, never\nre-staged, never straightened up, stood upright or otherwise re-posed.\nNo mannequin, grey figure or sketch line may remain anywhere in the\noutput.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 모닥불이 둥글게 타오르는 숲속 제단의 거대한 바위 앞에 강영란(흰 머리 여자)이 무릎을 꿇고 앉은 뒷모습.\n\nLOCATION (lock): In the ritual clearing before the huge boulder, encircled by burning campfires and drifting smoke. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The giant ritual rock remains stained with dark-red animal blood, and Youngran remains dressed in her red shaman’s robe.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S68sh1__bgfirst_bg.png"
    }
   ],
   "B": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S68sh1__bgfirst_bg.png"
    }
   ],
   "C": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S68sh1__bgfirst_bg.png"
    }
   ]
  },
  "totals": {
   "B": 7,
   "C": 6,
   "A": 5
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "레퍼런스의 구도를 완벽히 재현했으며, 흰 머리와 전통적인 붉은 무당복, 바위의 핏자국 등 모든 프롬프트 요소를 가장 정확하게 반영했습니다."
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "지시된 구도와 바위의 핏자국을 잘 표현했으나, 의상의 디자인이 전통 무당복보다는 다소 단순화된 형태입니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "구도와 배경은 레퍼런스와 일치하지만, 붉은 의상의 재질과 형태가 무당복이 아닌 현대적인 겨울 코트나 가운처럼 보입니다."
   }
  ],
  "refs": [
   {
    "label": "SHOT BACKGROUND",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S68sh1__bgfirst_bg.png"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S68sh1__bgfirst_bg.png",
   "bg_asset_id": "bcdee313-035b-4bbe-8cd6-0ac312f2c171",
   "bg_record_key": "S68sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "lane_conti_only"
  },
  "ref_mode": "lane(map_marker): 스케치+엔티티",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S68sh4": {
  "input_fingerprint": "8044b3ac739f4356",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)이 품속에서 꺼낸 짐승 뼈 모양의 목걸이를 허공으로 높이 치켜든 찰나.\n\nLOCATION (lock): At the altar directly before the giant boulder in the firelit forest clearing. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the ring of fires, giant blood-marked boulder, ritual smoke, and night forest. Exclude the elder's kneeling back pose and emphasize the raised bone pendant.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran has Jihu’s missing antler-bone pendant and raises it before placing it on the rock altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 4,
   "A": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지시문대로 뼈 목걸이를 강조하고 이전 샷의 뒷모습을 영리한 프레이밍으로 배제하여 배경과의 연속성을 완벽하게 유지했습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "목걸이가 아닌 거대한 동물 두개골을 들고 있으며, 이전 샷에서 무릎을 꿇고 있던 인물이 서 있는 모습으로 바뀌어 연속성에 어긋납니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "이전 샷의 비녀가 인물의 목을 가로질러 관통하는 치명적인 해부학적 오류(Hard Violation)가 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S68sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "손가락이 6개이며 구조가 기형적입니다.",
     "fix_en": "Correct the hand anatomy to have exactly five fingers, removing the extra pinky on the right and fixing the natural grip of the thumb and index finger on the string.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Correct the hand anatomy to have exactly five fingers, removing the extra pinky on the right and fixing the natural grip of the thumb and index finger on the string.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S68sh1"
  }
 },
 "S68sh7": {
  "input_fingerprint": "09841e5b25e4c30e",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)이 입술을 뾰족하게 오므리고 허공을 향해 날카롭게 휘파람을 부는 얼굴 클로즈업.\n\nLOCATION (lock): Before the boulder altar in the smoky, firelit ritual clearing at night. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the firelit ritual clearing, smoke, boulder, and red ceremonial clothing. Exclude the pendant-raising action and frame only the elder whistling.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains in Youngran’s possession at the ritual altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 5,
   "C": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지시된 휘파람 부는 표정과 클로즈업 프레이밍을 정확히 구현했으며, 이전 샷의 붉은 옷과 제단 바위 배경을 완벽하게 유지했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "표정과 프레이밍은 훌륭하나, 필수적으로 포함해야 할 배경의 제단 바위가 식별되지 않습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "표정과 배경은 유지되었으나, 불필요한 팔과 소매가 얼굴 옆으로 올라와 프레임을 가리고 있어 어색합니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S68sh4_sel.png"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S68sh4"
  }
 },
 "S69sh4::bgfirst_bg": {
  "input_fingerprint": "a7af841483012461",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 식인에 대한 진실을 깨닫고 경악하여 숲길 한가운데에 우뚝 멈춰 선 정인우의 전신.\n\nLOCATION (lock): In the middle of a dark pine-forest path illuminated only by a handheld flashlight.\n\nTIME OF DAY (lock): night, hazy sky.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at waist height several meters in front of 정인우, isolating his full body at the center of the forest path at the instant his forward movement stops. Keep the flashlight lowered within his silhouette and leave open path space around him so his realization reads as a physical rupture in the search.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 숲길 (수색 도중 멈춰 선 길); used as Provides the central corridor in which his halted body is isolated; 해송 나무들 (숲길 사이에 늘어서 있음) — Their trunks face the camera obliquely on both sides of the path; used as Creates vertical divisions around the lone figure and retains the search environment; 플래쉬 라이트 (켜져 있으며 흔들리다 멈춤) — Its front points low and away from the camera in 정인우's hand; used as Remains visible as the only carried search light before it is aimed toward 수리영; 희뿌연 연기 (나무 사이 밤하늘을 지나감); used as Passes through the visible night sky between the trees.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 식인에 대한 진실을 깨닫고 경악하여 숲길 한가운데에 우뚝 멈춰 선 정인우의 전신.\n\nLOCATION (lock): In the middle of a dark pine-forest path illuminated only by a handheld flashlight.\n\nTIME OF DAY (lock): night, hazy sky.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at waist height several meters in front of 정인우, isolating his full body at the center of the forest path at the instant his forward movement stops. Keep the flashlight lowered within his silhouette and leave open path space around him so his realization reads as a physical rupture in the search.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 숲길 (수색 도중 멈춰 선 길); used as Provides the central corridor in which his halted body is isolated; 해송 나무들 (숲길 사이에 늘어서 있음) — Their trunks face the camera obliquely on both sides of the path; used as Creates vertical divisions around the lone figure and retains the search environment; 플래쉬 라이트 (켜져 있으며 흔들리다 멈춤) — Its front points low and away from the camera in 정인우's hand; used as Remains visible as the only carried search light before it is aimed toward 수리영; 희뿌연 연기 (나무 사이 밤하늘을 지나감); used as Passes through the visible night sky between the trees.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S69sh4__bgfirst_bg.png",
  "asset_id": "1b8912bf-814f-4044-8590-5ced2c6c546e",
  "input_asset_ids": [
   "9ef5a060-1df8-43e4-86d7-aecec31432e4",
   "09cf0b8b-8857-49e2-af2b-fcea7222c63c"
  ]
 },
 "S69sh4": {
  "input_fingerprint": "acb0a8f855bb396a",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, hazy sky.\n\nSHOT TEXT (authoritative, Korean): 식인에 대한 진실을 깨닫고 경악하여 숲길 한가운데에 우뚝 멈춰 선 정인우의 전신.\n\nLOCATION (lock): In the middle of a dark pine-forest path illuminated only by a handheld flashlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s missing pendant remains at Youngran’s ritual altar; Inwoo’s injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, hazy sky.\n\nSHOT TEXT (authoritative, Korean): 식인에 대한 진실을 깨닫고 경악하여 숲길 한가운데에 우뚝 멈춰 선 정인우의 전신.\n\nLOCATION (lock): In the middle of a dark pine-forest path illuminated only by a handheld flashlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s missing pendant remains at Youngran’s ritual altar; Inwoo’s injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, hazy sky.\n\nSHOT TEXT (authoritative, Korean): 식인에 대한 진실을 깨닫고 경악하여 숲길 한가운데에 우뚝 멈춰 선 정인우의 전신.\n\nLOCATION (lock): In the middle of a dark pine-forest path illuminated only by a handheld flashlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s missing pendant remains at Youngran’s ritual altar; Inwoo’s injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S69sh4__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S69sh4.png"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1385833>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송숲_오솔길_22bdfa.png"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1385833>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 전신 샷 프레이밍과 손전등, 깁스 착용 상태를 정확히 구현했으며, 젊은 남성 캐릭터와 경악한 표정 연기도 훌륭하게 반영되었습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "캐릭터가 노년 여성의 얼굴로 잘못 묘사되었고, 손전등을 쥔 손과 깁스 주변의 해부학적 구조가 물리적으로 어색합니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 전신 샷 프레이밍과 손전등, 깁스 착용 상태를 정확히 구현했으며, 젊은 남성 캐릭터와 경악한 표정 연기도 훌륭하게 반영되었습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "캐릭터가 노년 여성의 얼굴로 잘못 묘사되었고, 손전등을 쥔 손과 깁스 주변의 해부학적 구조가 물리적으로 어색합니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "전신 샷과 숲길 배경은 잘 구현했으나, 지시된 정인우(청년)가 아닌 다른 인물(노년 여성)이 등장하여 캐릭터 일치도에서 크게 감점되었습니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "어두운 숲길 한가운데 멈춰 선 전신 샷의 구도를 잘 잡았으며, 깁스를 한 상태와 지시된 인물(정인우)의 외형 및 경악한 표정까지 충실히 구현했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "전신 샷과 숲길 배경은 잘 구현했으나, 지시된 정인우(청년)가 아닌 다른 인물(노년 여성)이 등장하여 캐릭터 일치도에서 크게 감점되었습니다."
     },
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "어두운 숲길 한가운데 멈춰 선 전신 샷의 구도를 잘 잡았으며, 깁스를 한 상태와 지시된 인물(정인우)의 외형 및 경악한 표정까지 충실히 구현했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 7
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 7
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지정된 전신 샷 프레이밍과 손전등, 깁스 착용 상태를 정확히 구현했으며, 젊은 남성 캐릭터와 경악한 표정 연기도 훌륭하게 반영되었습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "캐릭터가 노년 여성의 얼굴로 잘못 묘사되었고, 손전등을 쥔 손과 깁스 주변의 해부학적 구조가 물리적으로 어색합니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송숲_오솔길_22bdfa.png"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1385833>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "원본 배경 이미지를 그대로 유지해야 한다는 지침과 달리, 배경 상단에 안개가 추가되고 전체적으로 어둡게 변경되었습니다.",
     "fix_en": "Restore the exact original background image, removing the added fog and darkened vignette to keep the background's lighting and clarity completely untouched.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore the exact original background image, removing the added fog and darkened vignette to keep the background's lighting and clarity completely untouched.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S69sh4__bgfirst_bg.png",
   "bg_asset_id": "1b8912bf-814f-4044-8590-5ced2c6c546e",
   "bg_record_key": "S69sh4::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "해송숲_오솔길",
   "groupbg_asset_id": "09cf0b8b-8857-49e2-af2b-fcea7222c63c"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S69sh9": {
  "input_fingerprint": "b16f51b3a8a92901",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, hazy sky.\n\nSHOT TEXT (authoritative, Korean): 불이 꺼진 어두운 수풀 뒤로 김지후(강수리영)와 정인우가 바닥에 납작하게 엎드려 숨을 죽인 상태.\n\nLOCATION (lock): Behind low brush beside the unlit nighttime forest trail. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the pitch-dark pine forest, narrow path, brush, and extinguished flashlight state. Exclude the man's upright shocked stance and show both figures flattened behind foliage.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains at the ritual altar, and Inwoo’s injured arm remains cast as the pair hide in darkness.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 3,
   "C": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "수풀 뒤에 납작하게 엎드린 자세를 완벽히 연출했으며, 깁스한 왼팔과 복장 연속성도 정확히 일치합니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "엎드린 자세는 적절하나 남자의 깁스가 왼팔에서 오른팔로 뒤바뀐 치명적인 설정 오류가 있습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "깁스 위치와 의상은 맞으나 인물들이 바닥에 납작하게 엎드리지 않고 상체를 들고 있어 지시와 어긋납니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S69sh4_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1385833>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 샷에서 남자의 왼팔에 있던 깁스가 현재 이미지에서는 오른팔에 잘못 위치해 있습니다.",
     "fix_en": "Remove the blue cast from the man's right arm, restoring the green jacket sleeve and bare hand, and place the cast on his left arm.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the blue cast from the man's right arm, restoring the green jacket sleeve and bare hand, and place the cast on his left arm.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S69sh4"
  }
 },
 "S69sh10": {
  "input_fingerprint": "459446d9e1af4380",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, hazy sky.\n\nSHOT TEXT (authoritative, Korean): 칠흑 같은 어둠 속에서 등불 없이 산길을 바쁘게 내려가며 한쪽 무릎이 구부러진 금월도 마을 청년들의 기괴한 실루엣.\n\nLOCATION (lock): On a descending forest trail between dark tree trunks, with no lanterns or artificial light. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the dark path, pine trunks, undergrowth, and moonless ambience. Exclude the hidden pair and show only the youths descending without lamps.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s missing pendant remains at Youngran’s ritual altar.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "'한쪽 무릎이 구부러진 기괴한 자세'를 인물들의 실제 연기로 훌륭하게 구현했으며, 칠흑 같은 어둠 속 실루엣의 느낌과 지정된 배경 설정을 잘 유지했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "어두운 밤의 분위기와 배경 조건은 충족했으나, '한쪽 무릎이 구부러진 기괴한 실루엣'이라는 핵심적인 자세 지침을 간과하고 평범하게 걷는 모습으로 렌더링되었습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "'칠흑 같은 어둠'과 '실루엣'이라는 조명 지침을 무시하고 지나치게 밝게 렌더링되었으며, 기괴한 자세가 아닌 일반적인 달리기 동작으로 표현되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S69sh9_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:902182>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "가장 왼쪽에 있는 청년의 아래로 늘어뜨린 왼손이 손가락 구분이 없이 뭉개져 있습니다.",
     "fix_en": "Reconstruct the leftmost youth's hanging left hand so that the fingers are distinct and anatomically correct.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "오른쪽 체크무늬 셔츠를 입은 청년의 얼굴이 제공된 캐릭터 레퍼런스의 소년과 정확히 일치하지 않고 더 나이 들어 보입니다.",
     "fix_en": "Edit the face of the youth in the plaid shirt on the right to exactly match the specific boy in the character reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Reconstruct the leftmost youth's hanging left hand so that the fingers are distinct and anatomically correct.\n- Edit the face of the youth in the plaid shirt on the right to exactly match the specific boy in the character reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S69sh9"
  }
 },
 "S70sh1::bgfirst_bg": {
  "input_fingerprint": "7a8a6755c7cb71d2",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 기괴하게 나무 기둥이 덧대어진 거대한 자연동굴 입구 앞 어둠 속에 멈춰 선 김지후(강수리영)와 정인우의 뒷모습.\n\nLOCATION (lock): At the timber-braced mouth of the large natural cave in the nighttime forest.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at standing eye level several meters behind 김지후(강수리영) and 정인우, centering their backs beneath the full cave entrance. Place them apart across the lower middle of the wide frame so the supported opening and its dark interior remain the shared destination above and between them.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) and 정인우 in the lower-center of the frame, midground, looks toward cave interior; full cave entrance in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 거대한 자연동굴 입구 (입구가 열려 있음) — The opening faces the camera behind the two figures, revealing its dark interior; used as Primary environmental focal point enclosing the shared direction of both characters' attention; 덧대어진 나무 기둥 (동굴 입구에 기괴하게 덧대어짐) — Their supported sides and edges face the camera around the cave opening; used as Creates the entrance's strange reinforced outline and frames the darkness beyond; 동굴 내부 (짙은 어둠에 잠겨 있음); used as Provides negative space beyond the hesitant figures.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 기괴하게 나무 기둥이 덧대어진 거대한 자연동굴 입구 앞 어둠 속에 멈춰 선 김지후(강수리영)와 정인우의 뒷모습.\n\nLOCATION (lock): At the timber-braced mouth of the large natural cave in the nighttime forest.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at standing eye level several meters behind 김지후(강수리영) and 정인우, centering their backs beneath the full cave entrance. Place them apart across the lower middle of the wide frame so the supported opening and its dark interior remain the shared destination above and between them.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) and 정인우 in the lower-center of the frame, midground, looks toward cave interior; full cave entrance in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 거대한 자연동굴 입구 (입구가 열려 있음) — The opening faces the camera behind the two figures, revealing its dark interior; used as Primary environmental focal point enclosing the shared direction of both characters' attention; 덧대어진 나무 기둥 (동굴 입구에 기괴하게 덧대어짐) — Their supported sides and edges face the camera around the cave opening; used as Creates the entrance's strange reinforced outline and frames the darkness beyond; 동굴 내부 (짙은 어둠에 잠겨 있음); used as Provides negative space beyond the hesitant figures.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S70sh1__bgfirst_bg.png",
  "asset_id": "3b174573-9a87-49ae-94cf-0c2ffc88f6b0",
  "input_asset_ids": [
   "0a013a00-a770-45cc-8411-f058da71a506",
   "803cc956-9eff-49c4-b977-3e62ad55fc55"
  ]
 },
 "S70sh1": {
  "input_fingerprint": "6563b932a04ba52c",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 기괴하게 나무 기둥이 덧대어진 거대한 자연동굴 입구 앞 어둠 속에 멈춰 선 김지후(강수리영)와 정인우의 뒷모습.\n\nLOCATION (lock): At the timber-braced mouth of the large natural cave in the nighttime forest. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains at the ritual altar; Inwoo’s injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상); 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 기괴하게 나무 기둥이 덧대어진 거대한 자연동굴 입구 앞 어둠 속에 멈춰 선 김지후(강수리영)와 정인우의 뒷모습.\n\nLOCATION (lock): At the timber-braced mouth of the large natural cave in the nighttime forest. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains at the ritual altar; Inwoo’s injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상); 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 기괴하게 나무 기둥이 덧대어진 거대한 자연동굴 입구 앞 어둠 속에 멈춰 선 김지후(강수리영)와 정인우의 뒷모습.\n\nLOCATION (lock): At the timber-braced mouth of the large natural cave in the nighttime forest. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains at the ritual altar; Inwoo’s injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상); 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S70sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S70sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:903809>"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:903809>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "거대한 자연동굴과 기괴한 나무 기둥의 형태가 잘 표현되었으며, '어둠 속'이라는 조명과 두 인물의 뒷모습 구도를 충실히 구현했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "인물의 뒷모습과 깁스 등은 잘 반영되었으나, 너무 밝은 조명과 정형화된 갱도 형태의 구조물이 텍스트의 분위기와 다소 거리가 있습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "거대한 자연동굴과 기괴한 나무 기둥의 형태가 잘 표현되었으며, '어둠 속'이라는 조명과 두 인물의 뒷모습 구도를 충실히 구현했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "인물의 뒷모습과 깁스 등은 잘 반영되었으나, 너무 밝은 조명과 정형화된 갱도 형태의 구조물이 텍스트의 분위기와 다소 거리가 있습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "정인우의 깁스 상태와 조끼 의상 디테일을 훌륭하게 구현했으며, 기괴하게 덧대어진 거대한 동굴 입구의 묘사가 지문과 잘 일치합니다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "인물들의 뒷모습과 깁스 설정은 반영되었으나 정인우의 조끼 의상이 누락되었고, 동굴의 나무 구조물이 너무 정돈되어 있어 기괴한 느낌이 부족합니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "정인우의 깁스 상태와 조끼 의상 디테일을 훌륭하게 구현했으며, 기괴하게 덧대어진 거대한 동굴 입구의 묘사가 지문과 잘 일치합니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "인물들의 뒷모습과 깁스 설정은 반영되었으나 정인우의 조끼 의상이 누락되었고, 동굴의 나무 구조물이 너무 정돈되어 있어 기괴한 느낌이 부족합니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 10
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 10
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "거대한 자연동굴과 기괴한 나무 기둥의 형태가 잘 표현되었으며, '어둠 속'이라는 조명과 두 인물의 뒷모습 구도를 충실히 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "인물의 뒷모습과 깁스 등은 잘 반영되었으나, 너무 밝은 조명과 정형화된 갱도 형태의 구조물이 텍스트의 분위기와 다소 거리가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_pine_forest_route_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903809>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "정인우의 겉옷에서 레퍼런스 이미지에 있던 베이지색 다용도 조끼가 누락되었습니다.",
     "fix_en": "Add the beige multi-pocket vest over Inwoo's green jacket.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the beige multi-pocket vest over Inwoo's green jacket.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S70sh1__bgfirst_bg.png",
   "bg_asset_id": "3b174573-9a87-49ae-94cf-0c2ffc88f6b0",
   "bg_record_key": "S70sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "seed_bg"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S70sh3": {
  "input_fingerprint": "3bf5179655836741",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 동굴의 짙은 어둠 속으로 성큼 걷기 위해 한 발을 공중에 든 mid-stride 자세의 김지후(강수리영) 전신.\n\nLOCATION (lock): Across the cave threshold, stepping from the exterior forest into the pitch-dark interior passage. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the timber-braced cave mouth, surrounding pine forest, and deep black interior. Exclude the pair's earlier hesitation and show the woman stepping inside first.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains at the ritual altar, and Inwoo’s injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 2,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "동굴의 어둠 속으로 들어가는 방향과 한 발을 공중에 든 전신 자세를 명확하게 구현하여 프롬프트에 가장 충실합니다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "동굴 안이 아닌 밖으로 걸어 나오고 있으며, 지시된 공중에 뜬 발 자세가 생략되어 방향과 동작을 모두 위반했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "한 발을 든 자세는 표현되었으나, 동굴 어둠 속이 아닌 숲을 향해 밖으로 걷고 있어 핵심 걷기 방향을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S70sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "동굴 어둠 속으로 성큼 걷는다는 지문과 달리, 고개를 뒤로 돌려 바깥을 보고 있습니다.",
     "fix_en": "Turn her head to face forward into the cave interior, showing only the back of her head and long hair.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "공중에 들려 있는 왼쪽 발과 바닥 사이에 부자연스러운 검은 그림자 기둥이 이어져 있습니다.",
     "fix_en": "Erase the unnatural black shadow connecting the raised left foot to the ground, leaving clear space under the lifted boot.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Turn her head to face forward into the cave interior, showing only the back of her head and long hair.\n- Erase the unnatural black shadow connecting the raised left foot to the ground, leaving clear space under the lifted boot.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S70sh1"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S71sh2": {
  "input_fingerprint": "ffbe6856af906337",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 플래시 불빛에 의해 동굴 바닥에 흩어진 사람의 뼈와 뜯겨진 신체 부위들이 적나라하게 드러난 클로즈업.\n\nLOCATION (lock): On the bone-littered floor of the cave's dark entrance passage, revealed by an unstable flashlight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Severed hands, feet, heads, and scattered human bones are spread inertly across the cave floor, with the remains layered and piled throughout the passage. Jihu’s pendant remains at the ritual altar.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 8,
   "A": 4,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "플래시 불빛 아래 흩어진 뼈와 뜯겨진 신체 부위(머리, 팔, 발)를 클로즈업으로 정확하고 적나라하게 표현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "뼈와 밧줄, 천 조각 등만 보이며, 프롬프트에 명시된 잘린 신체 부위가 완전히 누락되었습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "요구된 잘린 신체 부위가 전혀 묘사되지 않았고, 현대적인 쓰레기(플라스틱 병, 운동화)가 주를 이룹니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L22B01.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "잘린 발의 발가락 개수가 6개 이상으로 해부학적으로 잘못되었습니다.",
     "fix_en": "Edit the severed foot to have exactly five anatomically correct toes.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "잘린 팔 끝에 달린 손의 손가락 개수가 6개 이상이며 기형적입니다.",
     "fix_en": "Edit the severed hand to have exactly five normal human fingers.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Edit the severed foot to have exactly five anatomically correct toes.\n- Edit the severed hand to have exactly five normal human fingers.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S71sh14": {
  "input_fingerprint": "6b78b849c5edfbdb",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 동굴 구석에서 두꺼운 쇠사슬에 묶인 채 잔뜩 웅크린 강민숙(홍은미)의 초췌한 전신이 플래시 불빛에 뚜렷이 드러난 광경.\n\nLOCATION (lock): In a deep corner of the cave beyond blood puddles and scattered bones, lit by a flashlight. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the rough cave walls, bone-strewn floor, intermittent flashlight, and deep darkness. Exclude the dismembered-remains close-up and reveal the chained woman in the corner.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is tightly curled on the floor in a corner of the cave with her torso folded forward, her head held low, and her limbs confined by thick chains.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk remains shackled in the cave; bones, body parts, blood pools, and wall spatters remain throughout it.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 6,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "이전 컷의 시체를 화면에서 배제하고, 지시문대로 플래시 조명이 고개를 숙인 채 웅크린 여성의 전신을 뚜렷하게 비추어 프롬프트를 가장 충실하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "여성이 고개를 푹 숙이고 있어야 한다는 고정 포즈 지침을 어기고 고개를 들어 앞을 바라보고 있어 감점이 불가피합니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "이전 컷의 시체를 배제하라는 지시와 달리 화면 전경에 배치했으며, 플래시 불빛 역시 여성이 아닌 시체를 향하고 있어 핵심 연출을 놓쳤습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S71sh2_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "왼쪽 팔을 잡고 있는 오른쪽 손의 손가락이 6개입니다.",
     "fix_en": "Remove the extra finger on the right hand gripping the left arm.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the extra finger on the right hand gripping the left arm.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S71sh2"
  }
 },
 "S71sh18": {
  "input_fingerprint": "715d5787448f83ba",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 눈물이 뺨을 타고 흘러내리며 입을 벌린 채 오열하는 표정으로 굳어 있는 강민숙(홍은미)의 얼굴 클로즈업.\n\nLOCATION (lock): At the chained holding spot deep within the bloodstained cave interior. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the cave's rough stone, flashlight lighting, chains, and captive position. Exclude the broad full-body reveal and focus on the captive woman's anguished face.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is tightly curled on the floor in a corner of the cave with her torso folded forward, her head held low, and her limbs confined by thick chains.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk remains with her limbs confined by thick chains amid the cave’s accumulated human remains and blood.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 5,
   "C": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "요구된 얼굴 클로즈업 앵글을 가장 잘 구현했으며, 눈물을 흘리며 오열하는 표정 지시를 충실히 따랐습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "표정 묘사는 적절하나 프레임이 너무 넓어져 얼굴 클로즈업 지시를 온전히 충족하지 못했습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "클로즈업 지시에 비해 불필요하게 신체가 많이 노출되었으며, 피사체의 시선이 다소 어색합니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S71sh14_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "캐릭터가 상체를 꼿꼿이 세우고 고개를 들어 정면을 보고 있어, 상체를 앞으로 숙이고 고개를 푹 숙인 레퍼런스의 지정된 자세와 어긋납니다.",
     "fix_en": "Tilt the woman's head downward and hunch her shoulders to restore the forward-folded, head-low posture from the reference.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "레퍼런스와 달리 목에 채워진 금속 고리가 우측 벽의 쇠사슬과 연결되어 있지 않고 끊어져 있습니다.",
     "fix_en": "Connect the metal collar around her neck to the thick rusty chain anchored to the rock wall on the right with a continuous chain.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Connect the metal collar around her neck to the thick rusty chain anchored to the rock wall on the right with a continuous chain.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S71sh14"
  }
 },
 "S72sh2::bgfirst_bg": {
  "input_fingerprint": "0f2a88c0f7f56b50",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 다 타버린 창고를 바라보며 허탈감에 빠져 멍하니 서 있는 황봉업(황 씨)의 측면.\n\nLOCATION (lock): In the restaurant's rear yard beside the completely burned storage shed, still releasing dark smoke.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At standing eye height on 황봉업(황 씨)'s side axis, the lateral track briefly settles around a medium distance to hold his vacant profile. He occupies the middle-right of the frame with slack shoulders and planted feet, while the burned warehouse remains beyond his line of sight on the left background.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 황봉업(황 씨) in the middle-right of the frame, midground, looks toward burned warehouse; burned warehouse in the middle-left of the frame, background.\n- KEY BACKGROUND ELEMENTS: burned warehouse (completely burned and emitting dark smoke) — Its burned exterior faces the camera beyond 황봉업(황 씨)'s profile; used as Visible destination of 황봉업(황 씨)'s fixed profile gaze; restaurant backyard (surrounding the warehouse at night); used as Environmental separation between the observer and the ruined structure.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 다 타버린 창고를 바라보며 허탈감에 빠져 멍하니 서 있는 황봉업(황 씨)의 측면.\n\nLOCATION (lock): In the restaurant's rear yard beside the completely burned storage shed, still releasing dark smoke.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At standing eye height on 황봉업(황 씨)'s side axis, the lateral track briefly settles around a medium distance to hold his vacant profile. He occupies the middle-right of the frame with slack shoulders and planted feet, while the burned warehouse remains beyond his line of sight on the left background.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 황봉업(황 씨) in the middle-right of the frame, midground, looks toward burned warehouse; burned warehouse in the middle-left of the frame, background.\n- KEY BACKGROUND ELEMENTS: burned warehouse (completely burned and emitting dark smoke) — Its burned exterior faces the camera beyond 황봉업(황 씨)'s profile; used as Visible destination of 황봉업(황 씨)'s fixed profile gaze; restaurant backyard (surrounding the warehouse at night); used as Environmental separation between the observer and the ruined structure.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S72sh2__bgfirst_bg.png",
  "asset_id": "14789c92-a033-479a-94c4-ec300932b183",
  "input_asset_ids": [
   "8a69a99b-ed47-44f5-afbc-a56714c9ab18",
   "f37a98eb-3ba8-4c79-8441-68cabae7cb90"
  ]
 },
 "S72sh2": {
  "input_fingerprint": "900a98c21d55bdc6",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 다 타버린 창고를 바라보며 허탈감에 빠져 멍하니 서 있는 황봉업(황 씨)의 측면.\n\nLOCATION (lock): In the restaurant's rear yard beside the completely burned storage shed, still releasing dark smoke. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The backyard shed is now burned out and continues to emit dark smoke.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 다 타버린 창고를 바라보며 허탈감에 빠져 멍하니 서 있는 황봉업(황 씨)의 측면.\n\nLOCATION (lock): In the restaurant's rear yard beside the completely burned storage shed, still releasing dark smoke. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The backyard shed is now burned out and continues to emit dark smoke.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 다 타버린 창고를 바라보며 허탈감에 빠져 멍하니 서 있는 황봉업(황 씨)의 측면.\n\nLOCATION (lock): In the restaurant's rear yard beside the completely burned storage shed, still releasing dark smoke. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The backyard shed is now burned out and continues to emit dark smoke.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S72sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S72sh2.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B06.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "레퍼런스의 공간 구조(창고, 가로등 위치)를 정확히 유지하며 인물의 허탈한 측면 모습을 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "인물과 배경의 배치를 위해 레퍼런스의 공간 구조와 비율을 크게 왜곡하여 장소 일치도가 떨어집니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "레퍼런스의 공간 구조(창고, 가로등 위치)를 정확히 유지하며 인물의 허탈한 측면 모습을 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "인물과 배경의 배치를 위해 레퍼런스의 공간 구조와 비율을 크게 왜곡하여 장소 일치도가 떨어집니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "레퍼런스의 장소와 불탄 창고의 디테일을 완벽하게 유지한 채 인물의 측면 구도를 정확히 연출했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "창고의 크기, 형태 및 주변 지형이 레퍼런스와 다르게 변형되어 장소 고정 조건을 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "레퍼런스의 장소와 불탄 창고의 디테일을 완벽하게 유지한 채 인물의 측면 구도를 정확히 연출했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "창고의 크기, 형태 및 주변 지형이 레퍼런스와 다르게 변형되어 장소 고정 조건을 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 7,
     "B": 15
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 7,
   "B": 15
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "레퍼런스의 공간 구조(창고, 가로등 위치)를 정확히 유지하며 인물의 허탈한 측면 모습을 훌륭하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "인물과 배경의 배치를 위해 레퍼런스의 공간 구조와 비율을 크게 왜곡하여 장소 일치도가 떨어집니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B06.png"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S72sh2__bgfirst_bg.png",
   "bg_asset_id": "14789c92-a033-479a-94c4-ec300932b183",
   "bg_record_key": "S72sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S72sh4": {
  "input_fingerprint": "950af507ebb85fa1",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어금니를 꽉 깨물며 끓어오르는 분노를 간신히 억누르는 황봉업(황 씨)의 살벌한 얼굴 클로즈업.\n\nLOCATION (lock): Beside the charred shed ruins in the nighttime rear yard of the restaurant. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the charred shed ruins, smoking timbers, dark backyard, and residual firelight. Exclude the wider standing pose and focus on the boatman's suppressed anger.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The shed remains a smoking, burned-out ruin.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 5,
   "A": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "요구된 '얼굴 클로즈업'에 가장 부합할 뿐만 아니라 레퍼런스의 자세, 복장, 배경을 매우 정확하게 유지했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "분노하는 표정과 레퍼런스와의 연속성은 좋으나, 샷이 다소 넓어 '얼굴 클로즈업'의 요건을 온전히 충족하지 못했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "얼굴 클로즈업보다는 미디엄 샷에 가깝고, 인물이 정면을 향해 있어 레퍼런스에서 설정된 시선 및 자세와 어긋납니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S72sh2_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지문에는 '어금니를 꽉 깨물며 간신히 억누르는'이라고 되어 있으나, 입을 벌리고 이를 드러내고 있어 억눌린 분노가 아닌 폭발하는 분노처럼 보입니다.",
     "fix_en": "Close the man's mouth completely, pressing his lips together and bulging his jaw muscles to show he is clenching his teeth and suppressing his anger.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Close the man's mouth completely, pressing his lips together and bulging his jaw muscles to show he is clenching his teeth and suppressing his anger.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S72sh2"
  }
 },
 "S73sh4::bgfirst_bg": {
  "input_fingerprint": "beaed2e2f7618d66",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 마을 청년 #2가 고산들의 귓가에 바짝 입을 대고 긴밀하게 속삭이는 순간.\n\nLOCATION (lock): At the edge of the firelit ritual clearing, behind the officiant and boulder altar.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At ear height within close conversational distance, the final portion of the dolly-in compresses the gap between 금월도 마을 청년's mouth on the left and 고산들's ear on the right. Their overlapping profiles share the foreground, while 고산들's tightening brow remains readable behind the whispering mouth rather than turning the moment into a frontal portrait.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 금월도 마을 청년 in the middle-left of the frame, foreground; 고산들 in the middle-right of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: bonfire (burning strongly); used as Soft background evidence of the active ritual space behind the intimate whisper; ritual rock (being circled during the ritual) — Only part of the side facing the gathered ritual space is visible behind them; used as Partially readable ritual context beyond the two compressed profiles.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 마을 청년 #2가 고산들의 귓가에 바짝 입을 대고 긴밀하게 속삭이는 순간.\n\nLOCATION (lock): At the edge of the firelit ritual clearing, behind the officiant and boulder altar.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At ear height within close conversational distance, the final portion of the dolly-in compresses the gap between 금월도 마을 청년's mouth on the left and 고산들's ear on the right. Their overlapping profiles share the foreground, while 고산들's tightening brow remains readable behind the whispering mouth rather than turning the moment into a frontal portrait.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 금월도 마을 청년 in the middle-left of the frame, foreground; 고산들 in the middle-right of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: bonfire (burning strongly); used as Soft background evidence of the active ritual space behind the intimate whisper; ritual rock (being circled during the ritual) — Only part of the side facing the gathered ritual space is visible behind them; used as Partially readable ritual context beyond the two compressed profiles.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S73sh4__bgfirst_bg.png",
  "asset_id": "8ad64330-ebe1-45ee-acbd-f7eaa5caa59c",
  "input_asset_ids": [
   "b06e1941-4613-4ffa-96a2-497411cd6b1c",
   "0c87a96e-b9dd-4f92-b65a-0091db3fe885"
  ]
 },
 "S73sh4": {
  "input_fingerprint": "fb299c8e9225b259",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 마을 청년 #2가 고산들의 귓가에 바짝 입을 대고 긴밀하게 속삭이는 순간.\n\nLOCATION (lock): At the edge of the firelit ritual clearing, behind the officiant and boulder altar. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains on the ritual altar.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 청년 (한국인 20~30대 남성, 짧은 검은 머리, 그을린 얼굴, 거친 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 마을 청년 #2가 고산들의 귓가에 바짝 입을 대고 긴밀하게 속삭이는 순간.\n\nLOCATION (lock): At the edge of the firelit ritual clearing, behind the officiant and boulder altar. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains on the ritual altar.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 청년 (한국인 20~30대 남성, 짧은 검은 머리, 그을린 얼굴, 거친 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 마을 청년 #2가 고산들의 귓가에 바짝 입을 대고 긴밀하게 속삭이는 순간.\n\nLOCATION (lock): At the edge of the firelit ritual clearing, behind the officiant and boulder altar. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant remains on the ritual altar.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 청년 (한국인 20~30대 남성, 짧은 검은 머리, 그을린 얼굴, 거친 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S73sh4__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S73sh4.png"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 청년: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:756328>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L23B04.png"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 청년: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:756328>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지문에서 요구한 '귓가에 바짝 입을 대고 속삭이는' 핵심 행동과 레퍼런스의 캐릭터 복장(비니, 장갑)을 매우 정확하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "입을 귀가 아닌 상대의 뺨과 코 부위에 대고 있어 속삭이는 행동 묘사가 어색하며, 청년의 비니가 누락되었습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지문에서 요구한 '귓가에 바짝 입을 대고 속삭이는' 핵심 행동과 레퍼런스의 캐릭터 복장(비니, 장갑)을 매우 정확하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "입을 귀가 아닌 상대의 뺨과 코 부위에 대고 있어 속삭이는 행동 묘사가 어색하며, 청년의 비니가 누락되었습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "마을 청년의 비니가 누락되었으며, 제단 뒤쪽 가장자리가 아닌 제단 바로 앞에 인물을 배치하여 위치 설정에 어긋납니다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "마을 청년의 복장(비니, 장갑 등)과 귓속말하는 행동이 정확히 일치하며, 배경을 통해 지정된 위치의 공간감을 잘 살렸습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "마을 청년의 비니가 누락되었으며, 제단 뒤쪽 가장자리가 아닌 제단 바로 앞에 인물을 배치하여 위치 설정에 어긋납니다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "마을 청년의 복장(비니, 장갑 등)과 귓속말하는 행동이 정확히 일치하며, 배경을 통해 지정된 위치의 공간감을 잘 살렸습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 8
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 15,
   "B": 8
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지문에서 요구한 '귓가에 바짝 입을 대고 속삭이는' 핵심 행동과 레퍼런스의 캐릭터 복장(비니, 장갑)을 매우 정확하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "입을 귀가 아닌 상대의 뺨과 코 부위에 대고 있어 속삭이는 행동 묘사가 어색하며, 청년의 비니가 누락되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L23B04.png"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 청년: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:756328>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "오른쪽 인물이 단일 캐릭터 레퍼런스와 다릅니다(다른 얼굴, 비니 미착용). 프롬프트 제약상 화면의 모든 인물은 제공된 레퍼런스와 동일해야 합니다.",
     "fix_en": "Replace the right person's head and clothing to exactly match the character reference, including the face, green beanie, and dark blue jacket.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "우측 전경의 원본 배경이 훼손되었습니다. 나무 기둥이 흙벽처럼 변형되었고 매달려 있던 사슴 뿔 장식이 사라졌습니다.",
     "fix_en": "Restore the original tree bark texture and the hanging deer antlers to the structure on the extreme right edge.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the right person's head and clothing to exactly match the character reference, including the face, green beanie, and dark blue jacket.\n- Restore the original tree bark texture and the hanging deer antlers to the structure on the extreme right edge.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S73sh4__bgfirst_bg.png",
   "bg_asset_id": "8ad64330-ebe1-45ee-acbd-f7eaa5caa59c",
   "bg_record_key": "S73sh4::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S73sh5": {
  "input_fingerprint": "1a4226a4913dce52",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 미간을 잔뜩 찌푸린 고산들이 제단의 강영란(흰 머리 여자)과 심각한 시선을 마주치고 있는 찰나.\n\nLOCATION (lock): Within the ritual clearing between the boulder altar and the watching villagers. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the burning ritual fires, giant boulder, musicians, smoke, and night forest. Exclude the messenger's whispering emphasis and show the large man exchanging a grave look with the elder.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains on the ritual altar while the burned shed remains ruined back at the restaurant.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 청년 (한국인 20~30대 남성, 짧은 검은 머리, 그을린 얼굴, 거친 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "지시된 캐릭터 레퍼런스의 외모(짧은 머리, 얼굴)를 정확히 반영했고 제단 위의 펜던트(인식표)를 포함했으며, 두 인물의 심각한 시선 교환을 완벽히 구현했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "두 인물의 시선 교환과 전반적인 분위기는 좋으나, 지정된 캐릭터 레퍼런스의 얼굴이 아닌 이전 샷의 인물 외모를 그대로 사용했고 제단의 펜던트가 누락되었습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "두 인물이 서로 시선을 마주치지 않아 샷 텍스트의 핵심 연출을 위반했으며, 캐릭터 레퍼런스 불일치 및 펜던트 누락 등 주요 지시사항을 놓쳤습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S73sh4_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 청년: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1264084>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 장면에서 왼쪽 남자(갈색 재킷)가 쓰고 있던 갈색 비니가 누락되고 맨머리로 묘사되었습니다.",
     "fix_en": "Add a brown ribbed beanie onto the head of the man on the left.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add a brown ribbed beanie onto the head of the man on the left.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S73sh4"
  }
 },
 "S74sh2::bgfirst_bg": {
  "input_fingerprint": "7b5ac7a71f73f210",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 강민숙(홍은미)이 몸을 뒤로 반쯤 돌린 채 분노에 찬 얼굴로 입을 크게 벌리고 있는 찰나.\n\nLOCATION (lock): On a dark forest escape route leading away from the cave.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level on 강민숙(홍은미)'s three-quarter front side, the dolly-in eases as she stops and catches the instant before completing her turn. Her half-rotated torso occupies the center-left midground, mouth open in anger toward 김지후(강수리영)'s off-screen position, with the forest path retained behind her to preserve the group's direction of travel.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: forest route (unlit at night); used as Spatial route behind the halted confrontation and the path she will resume descending; forest growth (being pushed through by the group) — The surrounding growth recedes behind both sides of her turning body; used as Layered side boundaries framing the abrupt half-turn without obscuring her face.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 강민숙(홍은미)이 몸을 뒤로 반쯤 돌린 채 분노에 찬 얼굴로 입을 크게 벌리고 있는 찰나.\n\nLOCATION (lock): On a dark forest escape route leading away from the cave.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level on 강민숙(홍은미)'s three-quarter front side, the dolly-in eases as she stops and catches the instant before completing her turn. Her half-rotated torso occupies the center-left midground, mouth open in anger toward 김지후(강수리영)'s off-screen position, with the forest path retained behind her to preserve the group's direction of travel.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: forest route (unlit at night); used as Spatial route behind the halted confrontation and the path she will resume descending; forest growth (being pushed through by the group) — The surrounding growth recedes behind both sides of her turning body; used as Layered side boundaries framing the abrupt half-turn without obscuring her face.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S74sh2__bgfirst_bg.png",
  "asset_id": "828ff95a-866b-4d8a-8093-674260eb5ffb",
  "input_asset_ids": [
   "fec68ab5-33de-41da-ab94-dfb84a376116",
   "09cf0b8b-8857-49e2-af2b-fcea7222c63c"
  ]
 },
 "S74sh2": {
  "input_fingerprint": "6bb1bf89d1ea3b6f",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 강민숙(홍은미)이 몸을 뒤로 반쯤 돌린 채 분노에 찬 얼굴로 입을 크게 벌리고 있는 찰나.\n\nLOCATION (lock): On a dark forest escape route leading away from the cave. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk is free of the cave chain, though the broken section and her remaining restraint hardware must match the escape. Jihu’s pendant remains at the altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 강민숙(홍은미)이 몸을 뒤로 반쯤 돌린 채 분노에 찬 얼굴로 입을 크게 벌리고 있는 찰나.\n\nLOCATION (lock): On a dark forest escape route leading away from the cave. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk is free of the cave chain, though the broken section and her remaining restraint hardware must match the escape. Jihu’s pendant remains at the altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 강민숙(홍은미)이 몸을 뒤로 반쯤 돌린 채 분노에 찬 얼굴로 입을 크게 벌리고 있는 찰나.\n\nLOCATION (lock): On a dark forest escape route leading away from the cave. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk is free of the cave chain, though the broken section and her remaining restraint hardware must match the escape. Jihu’s pendant remains at the altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S74sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S74sh2.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송숲_오솔길_22bdfa.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "몸을 반쯤 돌린 채 분노하며 입을 벌린 찰나의 표정과 자세를 정확히 연출했으며, 끊어진 사슬과 배경 조건도 충실히 반영했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "요구된 표정과 자세는 어느 정도 나타나나, 지시되지 않은 소품(손전등)과 옷 위의 부자연스러운 빛 번짐이 포함되어 몰입을 방해합니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "몸을 반쯤 돌린 채 분노하며 입을 벌린 찰나의 표정과 자세를 정확히 연출했으며, 끊어진 사슬과 배경 조건도 충실히 반영했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "요구된 표정과 자세는 어느 정도 나타나나, 지시되지 않은 소품(손전등)과 옷 위의 부자연스러운 빛 번짐이 포함되어 몰입을 방해합니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "레퍼런스의 어두운 밤 숲 분위기를 충실히 반영했으며, 몸을 반쯤 돌린 채 분노하여 입을 벌린 찰나의 표정과 남은 결박 도구의 상태를 지문대로 정확히 묘사했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "지문에 명시되지 않은 물건(왼손의 검은 원통형 물체)을 임의로 추가하여 감점되었으며, 복부 위치에 부자연스러운 빛 반사(렌즈 플레어) 오류가 존재합니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "레퍼런스의 어두운 밤 숲 분위기를 충실히 반영했으며, 몸을 반쯤 돌린 채 분노하여 입을 벌린 찰나의 표정과 남은 결박 도구의 상태를 지문대로 정확히 묘사했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "지문에 명시되지 않은 물건(왼손의 검은 원통형 물체)을 임의로 추가하여 감점되었으며, 복부 위치에 부자연스러운 빛 반사(렌즈 플레어) 오류가 존재합니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 9
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 9
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "몸을 반쯤 돌린 채 분노하며 입을 벌린 찰나의 표정과 자세를 정확히 연출했으며, 끊어진 사슬과 배경 조건도 충실히 반영했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "요구된 표정과 자세는 어느 정도 나타나나, 지시되지 않은 소품(손전등)과 옷 위의 부자연스러운 빛 번짐이 포함되어 몰입을 방해합니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송숲_오솔길_22bdfa.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "레이아웃 스케치 배경에 표시된 세 명의 인물이 누락되었습니다.",
     "fix_en": "Render the three small figures in the background path exactly at the positions and poses shown in the layout sketch.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Render the three small figures in the background path exactly at the positions and poses shown in the layout sketch.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S74sh2__bgfirst_bg.png",
   "bg_asset_id": "828ff95a-866b-4d8a-8093-674260eb5ffb",
   "bg_record_key": "S74sh2::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "해송숲_오솔길",
   "groupbg_asset_id": "09cf0b8b-8857-49e2-af2b-fcea7222c63c"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S74sh8": {
  "input_fingerprint": "903213a39687d697",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 숲 저편에서 날아든 휘파람 소리에 화들짝 놀란 표정으로 고개를 홱 돌린 채 굳어있는 김지후(강수리영)의 긴장된 얼굴.\n\nLOCATION (lock): Along the unlit forest path as a distant whistle echoes through the trees. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the dark pine path, hurried three-person formation, and moonlit foliage. Exclude the mother's angry turn and focus on the daughter reacting to the whistle.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains at the ritual altar; Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 5,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지문이 요구한 '긴장된 얼굴' 중심의 프레이밍을 가장 정확히 구현했으며, 이전 샷에서 설정된 배경 3인조와의 물리적 거리와 위치를 그대로 잘 유지했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "얼굴에 집중해야 하는 프레이밍 지시를 어기고 프레임을 넓혀 상반신과 팔을 과도하게 노출했으며, 배경의 3인조 중 1명이 누락되었습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "오직 카메라 위치만 변경되어야 한다는 지시를 어기고, 멀리 떨어져 있던 인물들을 여성의 바로 뒤로 이동시켜 공간적 연속성을 심각하게 훼손했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S74sh2_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "레퍼런스 이미지에 있던 어머니 캐릭터가 동일한 녹색 재킷과 헤어스타일로 그대로 등장했으며, 지시된 딸(김지후)의 모습으로 대체되지 않았습니다.",
     "fix_en": "Replace the foreground woman with a younger character (the daughter) wearing different clothing and a new hairstyle, completely removing the mother's green jacket.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the foreground woman with a younger character (the daughter) wearing different clothing and a new hairstyle, completely removing the mother's green jacket.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S74sh2"
  }
 },
 "S75sh2": {
  "input_fingerprint": "9ff4bab01b85cea3",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 동굴 바닥에 쓸쓸히 나뒹구는 끊어진 쇠사슬 단면의 클로즈업.\n\nLOCATION (lock): At the now-empty holding area deep inside the cave, lit by the searchers' torches. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The broken chain remains on the cave floor where Minsuk had been imprisoned; the space is now empty.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 6,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "끊어진 쇠사슬 단면을 완벽하게 클로즈업하여 샷의 핵심 구도를 가장 정확히 구현했습니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "프롬프트의 전반적인 분위기와 구도에 부합하나, 요구된 '단면'의 시각적 강조가 B에 비해 다소 약합니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "쇠사슬 고리들이 물리적으로 불가능한 형태로 바닥과 융합되어 있어 구조적 오류가 발생했습니다."
   }
  ],
  "refs": [],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S76sh1::bgfirst_bg": {
  "input_fingerprint": "1e8c0a3941a0e86f",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 입을 반쯤 벌린 채 숲길에 우뚝 멈춰 선 강민숙(홍은미)의 전신.\n\nLOCATION (lock): On the nighttime forest path beside the slaughterhouse and nearby goat pens.\n\nTIME OF DAY (lock): night, moonlight.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height on the path ahead of the group, the backward track settles into a static wide frame at the exact instant 강민숙(홍은미) arrests her stride. She occupies the middle-left in full figure with her body turned three-quarters toward the unseen attraction, while 김지후(강수리영) and 정인우 remain smaller and staggered behind her along the path.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 숲길 (Visible at night); used as The path supplies the receding axis that preserves the spacing between the abruptly halted leader and her companions; 숲 (Dark); used as The surrounding forest encloses both sides of the frame and withholds the object of 강민숙(홍은미)'s attention until the coming pivot.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 입을 반쯤 벌린 채 숲길에 우뚝 멈춰 선 강민숙(홍은미)의 전신.\n\nLOCATION (lock): On the nighttime forest path beside the slaughterhouse and nearby goat pens.\n\nTIME OF DAY (lock): night, moonlight.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height on the path ahead of the group, the backward track settles into a static wide frame at the exact instant 강민숙(홍은미) arrests her stride. She occupies the middle-left in full figure with her body turned three-quarters toward the unseen attraction, while 김지후(강수리영) and 정인우 remain smaller and staggered behind her along the path.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 숲길 (Visible at night); used as The path supplies the receding axis that preserves the spacing between the abruptly halted leader and her companions; 숲 (Dark); used as The surrounding forest encloses both sides of the frame and withholds the object of 강민숙(홍은미)'s attention until the coming pivot.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S76sh1__bgfirst_bg.png",
  "asset_id": "406c8acd-d14b-4772-af7d-d63c5d72749b",
  "input_asset_ids": [
   "d9b6eeb9-80c9-4a69-bf76-53e5b3606257",
   "0eb1234f-4ed6-464e-bf45-623129662d19"
  ]
 },
 "S76sh1": {
  "input_fingerprint": "ad2af2b3239dc3b1",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, moonlight.\n\nSHOT TEXT (authoritative, Korean): 입을 반쯤 벌린 채 숲길에 우뚝 멈춰 선 강민숙(홍은미)의 전신.\n\nLOCATION (lock): On the nighttime forest path beside the slaughterhouse and nearby goat pens. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains at the ritual altar; Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, moonlight.\n\nSHOT TEXT (authoritative, Korean): 입을 반쯤 벌린 채 숲길에 우뚝 멈춰 선 강민숙(홍은미)의 전신.\n\nLOCATION (lock): On the nighttime forest path beside the slaughterhouse and nearby goat pens. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains at the ritual altar; Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, moonlight.\n\nSHOT TEXT (authoritative, Korean): 입을 반쯤 벌린 채 숲길에 우뚝 멈춰 선 강민숙(홍은미)의 전신.\n\nLOCATION (lock): On the nighttime forest path beside the slaughterhouse and nearby goat pens. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains at the ritual altar; Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S76sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S76sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B03.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "레퍼런스의 구도를 피하면서 텍스트에 명시된 염소 우리를 반영해 앵글을 새롭게 잡았고, 입을 벌린 채 우뚝 선 전신의 모습이 정확히 연출되었습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "레퍼런스 이미지의 배경과 카메라 프레이밍을 그대로 복사하여 '프레이밍 복사 금지' 지침을 위반했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "레퍼런스의 구도를 피하면서 텍스트에 명시된 염소 우리를 반영해 앵글을 새롭게 잡았고, 입을 벌린 채 우뚝 선 전신의 모습이 정확히 연출되었습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "레퍼런스 이미지의 배경과 카메라 프레이밍을 그대로 복사하여 '프레이밍 복사 금지' 지침을 위반했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "입을 반쯤 벌린 전신 샷의 요구사항을 잘 충족하며, 레퍼런스 이미지의 쓰러진 나무와 지형 등 장소 디테일을 가장 정확하게 유지했습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "인물의 포즈와 표정은 적절하나, 레퍼런스 이미지 좌측에 없던 거대한 나무 구조물(우리)을 임의로 생성하여 장소 일치도가 떨어집니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "입을 반쯤 벌린 전신 샷의 요구사항을 잘 충족하며, 레퍼런스 이미지의 쓰러진 나무와 지형 등 장소 디테일을 가장 정확하게 유지했습니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "인물의 포즈와 표정은 적절하나, 레퍼런스 이미지 좌측에 없던 거대한 나무 구조물(우리)을 임의로 생성하여 장소 일치도가 떨어집니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 13
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 12,
   "B": 13
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "레퍼런스의 구도를 피하면서 텍스트에 명시된 염소 우리를 반영해 앵글을 새롭게 잡았고, 입을 벌린 채 우뚝 선 전신의 모습이 정확히 연출되었습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "레퍼런스 이미지의 배경과 카메라 프레이밍을 그대로 복사하여 '프레이밍 복사 금지' 지침을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B03.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "왼쪽 숲 배경에 참조 이미지에는 없는 나무 구조물(축사)이 임의로 추가되었습니다.",
     "fix_en": "Remove the wooden structure on the left and replace it with trees and dark foliage to match the reference photograph exactly.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "도축장 건물 입구 오른쪽에 열려 있는 철제 슬라이딩 문이 누락되었습니다.",
     "fix_en": "Restore the open, rusty metal door panel to the right side of the building's doorway, matching the architectural details of the reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the wooden structure on the left and replace it with trees and dark foliage to match the reference photograph exactly.\n- Restore the open, rusty metal door panel to the right side of the building's doorway, matching the architectural details of the reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S76sh1__bgfirst_bg.png",
   "bg_asset_id": "406c8acd-d14b-4772-af7d-d63c5d72749b",
   "bg_record_key": "S76sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S76sh5": {
  "input_fingerprint": "f1fae512eb43533e",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, moonlight.\n\nSHOT TEXT (authoritative, Korean): 도축장의 열린 문턱을 넘어 어둠 속으로 몸이 반쯤 진입한 채 앞으로 기울어진 자세의 강민숙(홍은미).\n\nLOCATION (lock): At the open slaughterhouse doorway, halfway from the moonlit forest path into the dark interior. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the moonlit path, slaughterhouse exterior, nearby goat enclosure, and darkness. Exclude the earlier stopped pose and show the woman entering through the open doorway.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains at the ritual altar; Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "레퍼런스의 도축장 내부 구조(타일, 호스 등)를 잘 유지하면서 문턱을 넘어 반쯤 진입해 기울어진 자세를 정확히 묘사했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "요구된 진입 자세는 무난하게 표현되었으나, 레퍼런스에 존재하던 도축장 내부의 구체적인 디테일이 대부분 생략되었습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "레퍼런스에 없던 외부 조명과 염소 우리를 임의로 크게 추가해 공간 설정을 위반했으며, 내부 구조도 완전히 묘사되지 않았습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S76sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "도축장 내부가 밝게 켜져 있어 '어둠 속으로' 진입한다는 지시와 모순됩니다.",
     "fix_en": "Darken the interior of the slaughterhouse completely to match the 'dark interior' setting."
    },
    {
     "issue_ko": "인물이 문턱을 완전히 넘어 내부에 위치해 있어 '몸이 반쯤 진입한 채'라는 지시를 위반했습니다.",
     "fix_en": "Redraw the character shifted slightly back so she is straddling the threshold, halfway outside."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Darken the interior of the slaughterhouse completely to match the 'dark interior' setting.\n- Redraw the character shifted slightly back so she is straddling the threshold, halfway outside.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S76sh1"
  }
 },
 "S77sh1::bgfirst_bg": {
  "input_fingerprint": "fddfb192957579f3",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 도축장 안에서 강민숙(홍은미)이 붉은 염소 피가 가득 든 양동이를 머리 위로 쏟아붓는 찰나.\n\nLOCATION (lock): Inside the dark forest slaughterhouse beside buckets filled with goat blood.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height near the entrance, a slow frontal dolly-in catches 강민숙(홍은미) at center as the raised bucket tips and goat blood begins to fall over her head and torso. The frame stays wide enough for 김지후(강수리영) and 정인우 to remain together at one side in the background, their recoiling bodies providing scale and reaction.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강민숙(홍은미) beneath the falling blood in the middle-center of the frame, midground; 김지후(강수리영) reacting at the side in the middle-right of the frame, background; 정인우 reacting beside her in the upper-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 도축장 내부 (Dark); used as The room provides sparse spatial context around the central action without distracting from the falling blood; 염소 피 양동이 (Being emptied) — Its open top is angled downward toward 강민숙(홍은미), with its side also visible to camera; used as The tipped container remains visible above her head and establishes the source of the blood; 붉은 염소 피 (Pouring); used as The falling blood connects the raised bucket to her head and torso as the principal action line.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 도축장 안에서 강민숙(홍은미)이 붉은 염소 피가 가득 든 양동이를 머리 위로 쏟아붓는 찰나.\n\nLOCATION (lock): Inside the dark forest slaughterhouse beside buckets filled with goat blood.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height near the entrance, a slow frontal dolly-in catches 강민숙(홍은미) at center as the raised bucket tips and goat blood begins to fall over her head and torso. The frame stays wide enough for 김지후(강수리영) and 정인우 to remain together at one side in the background, their recoiling bodies providing scale and reaction.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강민숙(홍은미) beneath the falling blood in the middle-center of the frame, midground; 김지후(강수리영) reacting at the side in the middle-right of the frame, background; 정인우 reacting beside her in the upper-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 도축장 내부 (Dark); used as The room provides sparse spatial context around the central action without distracting from the falling blood; 염소 피 양동이 (Being emptied) — Its open top is angled downward toward 강민숙(홍은미), with its side also visible to camera; used as The tipped container remains visible above her head and establishes the source of the blood; 붉은 염소 피 (Pouring); used as The falling blood connects the raised bucket to her head and torso as the principal action line.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S77sh1__bgfirst_bg.png",
  "asset_id": "d4f41ec9-7fbe-41ff-86d4-49e6efca77db",
  "input_asset_ids": [
   "f9c0c13d-f664-4c38-8f14-b48fc41e89c2",
   "c95571bc-a70d-45f1-96fd-710db821245b"
  ]
 },
 "S77sh1": {
  "input_fingerprint": "4d802c1253644423",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 도축장 안에서 강민숙(홍은미)이 붉은 염소 피가 가득 든 양동이를 머리 위로 쏟아붓는 찰나.\n\nLOCATION (lock): Inside the dark forest slaughterhouse beside buckets filled with goat blood. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk’s clothes, skin, and hair become thoroughly coated in goat blood to mask her scent.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강민숙(홍은미) (한국인 40대 후반 여성, 갸름한 얼굴, 단정한 검은 머리, 귀티 나는 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 도축장 안에서 강민숙(홍은미)이 붉은 염소 피가 가득 든 양동이를 머리 위로 쏟아붓는 찰나.\n\nLOCATION (lock): Inside the dark forest slaughterhouse beside buckets filled with goat blood. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk’s clothes, skin, and hair become thoroughly coated in goat blood to mask her scent.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강민숙(홍은미) (한국인 40대 후반 여성, 갸름한 얼굴, 단정한 검은 머리, 귀티 나는 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 도축장 안에서 강민숙(홍은미)이 붉은 염소 피가 가득 든 양동이를 머리 위로 쏟아붓는 찰나.\n\nLOCATION (lock): Inside the dark forest slaughterhouse beside buckets filled with goat blood. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk’s clothes, skin, and hair become thoroughly coated in goat blood to mask her scent.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강민숙(홍은미) (한국인 40대 후반 여성, 갸름한 얼굴, 단정한 검은 머리, 귀티 나는 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S77sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S77sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:844015>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L26B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:844015>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지문에 지정된 인물만 등장하여 양동이로 피를 붓는 동작과 장소의 분위기를 정확히 연출했습니다."
     },
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "지문에 언급되지 않은 두 명의 인물이 배경에 추가되어 심각한 규칙 위반(Hard Violation)입니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지문에 지정된 인물만 등장하여 양동이로 피를 붓는 동작과 장소의 분위기를 정확히 연출했습니다."
     },
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "지문에 언급되지 않은 두 명의 인물이 배경에 추가되어 심각한 규칙 위반(Hard Violation)입니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "지정된 단일 인물만 등장하여 규칙을 준수했으나, 피를 머리 위가 아닌 어깨와 가슴 쪽으로 쏟아붓고 있어 동작의 정확도가 다소 떨어집니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "피를 머리 위로 붓는 동작은 잘 표현되었으나, 프롬프트에 명시되지 않은 두 명의 인물을 배경에 임의로 추가하여 심각한 지침 위반에 해당합니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "지정된 단일 인물만 등장하여 규칙을 준수했으나, 피를 머리 위가 아닌 어깨와 가슴 쪽으로 쏟아붓고 있어 동작의 정확도가 다소 떨어집니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "피를 머리 위로 붓는 동작은 잘 표현되었으나, 프롬프트에 명시되지 않은 두 명의 인물을 배경에 임의로 추가하여 심각한 지침 위반에 해당합니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 5,
     "B": 13
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 5,
   "B": 13
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지문에 지정된 인물만 등장하여 양동이로 피를 붓는 동작과 장소의 분위기를 정확히 연출했습니다."
   },
   {
    "label": "A",
    "score": 2,
    "verdict_ko": "지문에 언급되지 않은 두 명의 인물이 배경에 추가되어 심각한 규칙 위반(Hard Violation)입니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L26B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:844015>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "강민숙이 양동이를 머리 위가 아닌 왼쪽 어깨 부근에서 쏟아붓고 있습니다.",
     "fix_en": "Shift the bucket and her hands so the bucket is directly over her head, with the blood pouring down onto the crown of her head.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "캐릭터 레퍼런스의 의상(카멜색 코트, 흰색 터틀넥, 스카프)과 일치하지 않고 임의의 어두운 재킷을 입고 있습니다.",
     "fix_en": "Replace her jacket with the camel coat, white turtleneck, and patterned scarf from the character reference, keeping the blood stains.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "머리카락과 피부가 피로 완전히 뒤덮여야 한다는 지시와 달리 머리와 얼굴 윗부분이 피가 묻지 않고 깨끗합니다.",
     "fix_en": "Cover her hair, face, and exposed skin completely with a thick coating of dark red blood.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Shift the bucket and her hands so the bucket is directly over her head, with the blood pouring down onto the crown of her head.\n- Replace her jacket with the camel coat, white turtleneck, and patterned scarf from the character reference, keeping the blood stains.\n- Cover her hair, face, and exposed skin completely with a thick coating of dark red blood.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S77sh1__bgfirst_bg.png",
   "bg_asset_id": "d4f41ec9-7fbe-41ff-86d4-49e6efca77db",
   "bg_record_key": "S77sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S77sh8": {
  "input_fingerprint": "9a6830c8f488b3ed",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 피가 든 양동이를 정인우의 몸을 향해 확 들이붓는 순간.\n\nLOCATION (lock): The slaughterhouse work area beside the blood buckets and butchering equipment. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: The forward move ends at upper-torso height from a three-quarter angle exactly as 김지후(강수리영) snaps the bucket toward 정인우. 김지후 drives diagonally from the left midground while 정인우 recoils on the right, and the pouring arc bridges their shared space without either figure leaving the frame.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, midground, reaches for 정인우 with the tipped bucket; 정인우 in the middle-right of the frame, midground.\n- KEY BACKGROUND ELEMENTS: bucket of goat blood (tipped and pouring) — Its opening angles from 김지후(강수리영) toward 정인우's upper body; used as Connects the two figures through the direction of the sudden pour; slaughterhouse interior (dark); used as Remains subdued behind the two-person action axis.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key interior lighting preserves strong contrast, with the red goat blood cutting across the restrained dark palette.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same dark slaughterhouse, blood buckets, hanging tools, and the mother's already blood-soaked state. Exclude her pouring action; show the daughter throwing blood over the man.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk remains drenched in goat blood as Jihu pours another bucket over Inwoo; his injured arm is still casted.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 김지후(강수리영) right now, so 김지후(강수리영)'s hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 김지후(강수리영): its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face); 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 3,
   "B": 6,
   "C": 8
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "붓는 방향과 착지는 맞지만 샷 텍스트에 없는 제3의 인물이 프레임 중앙에 추가된 결정적 위반이며, 지후의 피에 젖은 상태 계승도 어긋난다.  ★위반: [claude-opus] 샷 텍스트가 지정하지 않은 제3의 인물(피에 젖은 여성)이 프레임 중앙에 추가됨 — 발명된 인물 / [gpt] 김지후와 정인우 외에 피에 젖은 세 번째 인물이 추가되어 있다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "2인 액션 축과 붓는 방향·지지 자세는 정확하나, 인우의 팔에 민석의 것이어야 할 깁스가 붙어 소품 귀속이 틀렸고 지후 얼굴·헤어가 캐릭터 레퍼런스에서 크게 벗어난다.  ★위반: [gpt] 정인우에게 지정되지 않은 흰 손목 깁스가 추가되어 있으며, 민석의 부상 상태가 잘못된 인물에게 전이되었다."
   },
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "중좌 지후·중우 인우의 미디엄 3/4 구도, 양동이 입구에서 인우 상반신으로 이어지는 붓는 호, 레퍼런스 도축장 고정 요소와 젖은 상태 계승이 모두 맞아 가장 충실하다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S77sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:813104>"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:787232>"
   },
   {
    "label": "PROP REFERENCE — 염소 피 양동이: the exact object appearing in this shot; match its look, material and wear exactly.",
    "path": "<bytes:1175562>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "[Critical] 프롬프트는 시간대를 '밤(night)'으로 고정했으나, 왼쪽 창문들을 통해 밝은 낮의 햇빛이 들어오고 있습니다.",
     "fix_en": "Darken the left windows to completely black to reflect night time, removing the bright daylight shining through them. Preserve the characters Jihu and Inwoo, their poses, clothing, the blood, the bucket, the interior room lighting, and the framing exactly as they are."
    },
    {
     "issue_ko": "[Critical] 양동이에서 정인우에게 쏟아지는 피가 사실적인 액체가 아니라, 표면이 매끄러운 붉은색 고체 관(유리나 플라스틱)처럼 묘사되었습니다.",
     "fix_en": "Replace the solid, smooth red arc connecting the bucket to Inwoo with a turbulent stream of liquid blood, adding chaotic splashes, flying droplets, and a realistic liquid texture. Preserve Jihu, Inwoo, their poses, clothing, the bucket, the background, the lighting, and the framing exactly as they are."
    },
    {
     "issue_ko": "[Critical] 정인우의 다친 팔에 깁스가 있어야 한다는 설정이 누락되었으며, 양팔 모두 코트 소매만 보입니다.",
     "fix_en": "Add a white medical plaster cast to Inwoo's left forearm, showing it protruding from his green coat sleeve. Preserve Inwoo's face, pose, green coat, Jihu, her pose, clothing, the blood, the bucket, the background, the lighting, and the framing exactly as they are."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 5,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Darken the left windows to completely black to reflect night time, removing the bright daylight shining through them. Preserve the characters Jihu and Inwoo, their poses, clothing, the blood, the bucket, the interior room lighting, and the framing exactly as they are.\n- Replace the solid, smooth red arc connecting the bucket to Inwoo with a turbulent stream of liquid blood, adding chaotic splashes, flying droplets, and a realistic liquid texture. Preserve Jihu, Inwoo, their poses, clothing, the bucket, the background, the lighting, and the framing exactly as they are.\n- Add a white medical plaster cast to Inwoo's left forearm, showing it protruding from his green coat sleeve. Preserve Inwoo's face, pose, green coat, Jihu, her pose, clothing, the blood, the bucket, the background, the lighting, and the framing exactly as they are.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "구도·장소·인물은 맞지만 피가 매끈한 고무 리본처럼 굳어 액체 물성 조항을 어기고, CARRIED STATE가 요구한 정인우의 깁스한 팔이 없다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일 구도를 유지한 채 피를 실제 비말·분산으로 되돌리고 요구된 깁스 팔까지 복원했다; 좌측 창문이 다소 어두워진 것 외에 잃은 것이 거의 없어 수리의 이득이 비용을 넘는다.  ★위반: [gpt] 정인우의 오른쪽 전완에 계약이 지정하지 않은 흰 깁스를 새로 추가해 다른 인물인 민석의 소지 상태를 잘못 이전했다. / [gpt] 이전 스틸의 왼쪽 창을 가리는 크고 평평한 검은 사각 패널을 추가하여 잠긴 도축장 구조에 없는 물체·편집 흔적을 만들었다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.75,
      "B": 1.571
     },
     "adjusted": {
      "A": 1.75,
      "B": 1.071
     },
     "violations": {
      "B": [
       "[gpt] 정인우의 오른쪽 전완에 계약이 지정하지 않은 흰 깁스를 새로 추가해 다른 인물인 민석의 소지 상태를 잘못 이전했다.",
       "[gpt] 이전 스틸의 왼쪽 창을 가리는 크고 평평한 검은 사각 패널을 추가하여 잠긴 도축장 구조에 없는 물체·편집 흔적을 만들었다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "A"
     },
     "agreed": false
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "구도·장소·인물은 맞지만 피가 매끈한 고무 리본처럼 굳어 액체 물성 조항을 어기고, CARRIED STATE가 요구한 정인우의 깁스한 팔이 없다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일 구도를 유지한 채 피를 실제 비말·분산으로 되돌리고 요구된 깁스 팔까지 복원했다; 좌측 창문이 다소 어두워진 것 외에 잃은 것이 거의 없어 수리의 이득이 비용을 넘는다.  ★위반: [gpt] 정인우의 오른쪽 전완에 계약이 지정하지 않은 흰 깁스를 새로 추가해 다른 인물인 민석의 소지 상태를 잘못 이전했다. / [gpt] 이전 스틸의 왼쪽 창을 가리는 크고 평평한 검은 사각 패널을 추가하여 잠긴 도축장 구조에 없는 물체·편집 흔적을 만들었다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "왼쪽 중경의 여성이 양동이를 오른쪽으로 확 기울여 입구가 남자 상반신을 향하고, 피는 실제로 남자의 가슴·어깨·목에 도달해 코트에 붉게 튄다. 여성 시선은 남자 얼굴/상체를 겨냥, 남자는 시선을 아래로 피하며 몸을 뒤로 젖힌다. 조준 대상 = 정인우 상반신, 일치.",
      "built_space": "도축장 내부: 판자벽, 천장 전구 1개, 천장 훅 열, 오른쪽 벽 칼걸이, 중앙 나무 작업대, 하단 오른쪽 피 양동이 무리 — 이전 스틸의 고정 요소가 개수·배치 모두 유지. 왼쪽 창문 부분이 어둡게 눌려 있음(참조엔 흐린 창유리). 양동이는 화면 내 1개(들린 것) + 배경 양동이들, 중복 없음.",
      "entities": "여성: 한국인 20대 초반, 검은 머리 묶음, 피에 젖은 어두운 재킷·바지 — 이전 스틸의 피범벅 상태 유지(단 참조 캐릭터의 캡은 없음, 이전 스틸에도 없음). 남성: 한국인 20대 중반, 짧은 검은 머리, 녹색 울 코트+버건디 터틀넥 — 정인우 레퍼런스와 일치. 오른팔에 흰 깁스 붕대 노출 = 계약의 '깁스된 팔' 충족. 양동이는 아연도금 금속에 녹/피 자국, 나무 손잡이 계열 — 프롭 레퍼런스와 부합.",
      "hard_violations": [],
      "physics": "여성은 두 발이 바닥에 있고 앞쪽 다리에 체중을 실어 상체를 앞으로 던지는 자세, 두 손 모두 양동이 몸통과 테두리를 잡고 있음(그립 확인됨). 튄 피는 양동이 입구에서 분산된 물보라 형태로 관성 방향과 일치하며 남자 몸에 착지, 이미 코트에 젖은 자국이 생김. 공중에 지지 없는 물체 없음. 액체 표면이 광택 있고 방울 분리가 자연스러움."
     },
     {
      "label": "B",
      "direction": "동일 구도, 양동이 입구는 남자 쪽을 향하고 피 덩어리도 남자 목·가슴에 닿는다. 조준 자체는 맞음. 다만 피가 하나의 굵은 아치형 튜브로 남자 목에 꽂히듯 이어져, 순간적 '확 들이붓는' 비산감보다 정지된 리본처럼 읽힌다.",
      "built_space": "동일한 도축장: 판자벽, 전구, 훅, 칼걸이, 작업대, 하단 양동이 무리 유지. 왼쪽 창문이 A보다 밝게 드러나 이전 스틸의 흐린 유리창에 더 가깝다. 구조 중복 없음.",
      "entities": "여성: A와 동일 인물·의상·피범벅 상태. 남성: 녹색 코트·버건디 터틀넥, 정인우 레퍼런스 일치. 그러나 오른팔의 흰 깁스가 사라지고 맨 손목만 보임 — CARRIED STATE의 '깁스된 팔' 항목이 화면에서 소실. 코트에 묻은 피량도 A보다 적다.",
      "hard_violations": [],
      "physics": "여성 발과 그립은 A와 동일하게 지면·양동이에 정상 지지. 문제는 액체: 피가 두께 일정한 매끈한 아치 형태로 공중에 고정되어 있고, 낙하 곡선·분산·물방울이 없어 사탕/플라스틱 조형물처럼 읽힌다. 중력 하강이 거의 없는 이 덩어리를 지지하는 것이 없으며, MATERIAL REALISM(액체는 광택 있게 튀고 고인다)에 어긋난다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "양동이 입구가 정인우 상반신을 정확히 겨냥하고 피가 실제로 몸에 튀어 착지하며, 깁스된 팔과 어머니의 피범벅 상태까지 계약대로 유지 — 3/4 앵글 상반신 높이 미디엄샷 구성도 충족."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "구도·위치는 같으나 피가 중력 없는 균일한 아치 덩어리로 굳어 액체 물성(MATERIAL REALISM)을 잃었고, 계약이 요구한 '깁스된 팔'까지 화면에서 사라져 얻은 것보다 잃은 것이 크다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "양동이 입구가 정인우 상반신을 정확히 겨냥하고 피가 실제로 몸에 튀어 착지하며, 깁스된 팔과 어머니의 피범벅 상태까지 계약대로 유지 — 3/4 앵글 상반신 높이 미디엄샷 구성도 충족."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "구도·위치는 같으나 피가 중력 없는 균일한 아치 덩어리로 굳어 액체 물성(MATERIAL REALISM)을 잃었고, 계약이 요구한 '깁스된 팔'까지 화면에서 사라져 얻은 것보다 잃은 것이 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "왼쪽 중경의 여성이 양동이를 오른쪽으로 확 기울여 입구가 남자 상반신을 향하고, 피는 실제로 남자의 가슴·어깨·목에 도달해 코트에 붉게 튄다. 여성 시선은 남자 얼굴/상체를 겨냥, 남자는 시선을 아래로 피하며 몸을 뒤로 젖힌다. 조준 대상 = 정인우 상반신, 일치.",
      "built_space": "도축장 내부: 판자벽, 천장 전구 1개, 천장 훅 열, 오른쪽 벽 칼걸이, 중앙 나무 작업대, 하단 오른쪽 피 양동이 무리 — 이전 스틸의 고정 요소가 개수·배치 모두 유지. 왼쪽 창문 부분이 어둡게 눌려 있음(참조엔 흐린 창유리). 양동이는 화면 내 1개(들린 것) + 배경 양동이들, 중복 없음.",
      "entities": "여성: 한국인 20대 초반, 검은 머리 묶음, 피에 젖은 어두운 재킷·바지 — 이전 스틸의 피범벅 상태 유지(단 참조 캐릭터의 캡은 없음, 이전 스틸에도 없음). 남성: 한국인 20대 중반, 짧은 검은 머리, 녹색 울 코트+버건디 터틀넥 — 정인우 레퍼런스와 일치. 오른팔에 흰 깁스 붕대 노출 = 계약의 '깁스된 팔' 충족. 양동이는 아연도금 금속에 녹/피 자국, 나무 손잡이 계열 — 프롭 레퍼런스와 부합.",
      "hard_violations": [],
      "physics": "여성은 두 발이 바닥에 있고 앞쪽 다리에 체중을 실어 상체를 앞으로 던지는 자세, 두 손 모두 양동이 몸통과 테두리를 잡고 있음(그립 확인됨). 튄 피는 양동이 입구에서 분산된 물보라 형태로 관성 방향과 일치하며 남자 몸에 착지, 이미 코트에 젖은 자국이 생김. 공중에 지지 없는 물체 없음. 액체 표면이 광택 있고 방울 분리가 자연스러움."
     },
     {
      "label": "A",
      "direction": "동일 구도, 양동이 입구는 남자 쪽을 향하고 피 덩어리도 남자 목·가슴에 닿는다. 조준 자체는 맞음. 다만 피가 하나의 굵은 아치형 튜브로 남자 목에 꽂히듯 이어져, 순간적 '확 들이붓는' 비산감보다 정지된 리본처럼 읽힌다.",
      "built_space": "동일한 도축장: 판자벽, 전구, 훅, 칼걸이, 작업대, 하단 양동이 무리 유지. 왼쪽 창문이 A보다 밝게 드러나 이전 스틸의 흐린 유리창에 더 가깝다. 구조 중복 없음.",
      "entities": "여성: A와 동일 인물·의상·피범벅 상태. 남성: 녹색 코트·버건디 터틀넥, 정인우 레퍼런스 일치. 그러나 오른팔의 흰 깁스가 사라지고 맨 손목만 보임 — CARRIED STATE의 '깁스된 팔' 항목이 화면에서 소실. 코트에 묻은 피량도 A보다 적다.",
      "hard_violations": [],
      "physics": "여성 발과 그립은 A와 동일하게 지면·양동이에 정상 지지. 문제는 액체: 피가 두께 일정한 매끈한 아치 형태로 공중에 고정되어 있고, 낙하 곡선·분산·물방울이 없어 사탕/플라스틱 조형물처럼 읽힌다. 중력 하강이 거의 없는 이 덩어리를 지지하는 것이 없으며, MATERIAL REALISM(액체는 광택 있게 튀고 고인다)에 어긋난다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S77sh1"
  }
 },
 "S77sh10": {
  "input_fingerprint": "18d45df28e897ed6",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 재빨리 다른 염소 피 양동이를 자신의 머리 위로 쏟아붓는 찰나.\n\nLOCATION (lock): Inside the nighttime slaughterhouse at the blood-bucket area. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same slaughterhouse interior, blood buckets, rough timber, and night lighting. Exclude the blood pour over the man and show the daughter dousing herself.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Inwoo, and Jihu are all thoroughly coated in goat blood; Jihu’s pendant remains at the altar.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 4,
   "A": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "염소 피를 쏟아붓는 행동과 인물 단독 구도를 잘 구현했으나, 제단에 두어야 할 펜던트를 여전히 착용하고 있어 약간의 감점이 있습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "지시된 염소 피가 아닌 투명한 물을 뒤집어쓰고 있어 프롬프트의 핵심 시각적 요소를 크게 위반했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "텍스트에 명시되지 않은 남성 인물이 프레임에 포함되었으며, 양동이를 든 팔의 해부학적 구조가 불가능하게 왜곡된 하드 위반이 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S77sh8_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지후가 자신의 머리 위로 피를 쏟아붓는다는 설정과 달리, 양동이에서 나오는 핏줄기가 그녀의 머리를 완전히 빗겨가 오른쪽 허공으로 쏟아지고 있습니다.",
     "fix_en": "Redirect the main stream of blood from the bucket so it pours directly onto the top of Jihu's head and cascades down her body.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "화면 우측의 팔이 스웨터를 입지 않은 맨팔로 잘못 그려져 있으며, 찢겨나간 듯한 초록색 소매 조각들이 어깨와 양동이 쪽에 분리되어 떠 있습니다.",
     "fix_en": "Clothe the exposed left arm (on the right side of the frame) in the green cable-knit sweater sleeve to match the other arm, and remove the detached sleeve fragments from the shoulder and bucket.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "지후의 목걸이는 제단에 남아있어야 한다는 지시사항(CARRIED STATE)과 달리 목걸이를 착용하고 있습니다.",
     "fix_en": "Remove the pendant and its dark strings from Jihu's chest and neck.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redirect the main stream of blood from the bucket so it pours directly onto the top of Jihu's head and cascades down her body.\n- Clothe the exposed left arm (on the right side of the frame) in the green cable-knit sweater sleeve to match the other arm, and remove the detached sleeve fragments from the shoulder and bucket.\n- Remove the pendant and its dark strings from Jihu's chest and neck.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S77sh8"
  }
 },
 "S78sh1::bgfirst_bg": {
  "input_fingerprint": "16698af2f84e0c1c",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 일렁이는 모닥불 불빛과 핏빛 바위 앞에서 강영란(흰 머리 여자)이 기괴하고 날 선 표정으로 서 있는 전경.\n\nLOCATION (lock): Before the blood-smeared boulder in the forest ritual clearing, lit by flickering campfires.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level and medium-long distance, a static frontal composition places 강영란(흰 머리 여자) at center with the fire low in the near foreground and the blood-colored rock rising behind her. 고산들 appears only as a dark over-shoulder figure at the lower edge, turning the direct view of her sharpened expression into his confronted position.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강영란(흰 머리 여자) before the rock in the middle-center of the frame, midground; 고산들 confronting her in the lower-right of the frame, foreground; blood-colored rock in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 모닥불 (Lit and flickering); used as The fire occupies a low foreground band and visually agitates the otherwise static confrontation; 핏빛 바위 (Visible behind 강영란(흰 머리 여자)); used as The rock forms the central background mass directly behind 강영란(흰 머리 여자).\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 일렁이는 모닥불 불빛과 핏빛 바위 앞에서 강영란(흰 머리 여자)이 기괴하고 날 선 표정으로 서 있는 전경.\n\nLOCATION (lock): Before the blood-smeared boulder in the forest ritual clearing, lit by flickering campfires.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level and medium-long distance, a static frontal composition places 강영란(흰 머리 여자) at center with the fire low in the near foreground and the blood-colored rock rising behind her. 고산들 appears only as a dark over-shoulder figure at the lower edge, turning the direct view of her sharpened expression into his confronted position.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강영란(흰 머리 여자) before the rock in the middle-center of the frame, midground; 고산들 confronting her in the lower-right of the frame, foreground; blood-colored rock in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 모닥불 (Lit and flickering); used as The fire occupies a low foreground band and visually agitates the otherwise static confrontation; 핏빛 바위 (Visible behind 강영란(흰 머리 여자)); used as The rock forms the central background mass directly behind 강영란(흰 머리 여자).\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S78sh1__bgfirst_bg.png",
  "asset_id": "34feb538-20dc-4dd5-81eb-31cb517203ed",
  "input_asset_ids": [
   "9536fecf-a083-4e12-979b-788e985800a0",
   "025ebc30-17e5-432f-b444-8e54016fce74"
  ]
 },
 "S78sh1": {
  "input_fingerprint": "4e6be1dde83f623f",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 일렁이는 모닥불 불빛과 핏빛 바위 앞에서 강영란(흰 머리 여자)이 기괴하고 날 선 표정으로 서 있는 전경.\n\nLOCATION (lock): Before the blood-smeared boulder in the forest ritual clearing, lit by flickering campfires. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The ritual rock remains bloodstained, Youngran remains in her red shaman’s robe, and Jihu’s pendant remains on the altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 일렁이는 모닥불 불빛과 핏빛 바위 앞에서 강영란(흰 머리 여자)이 기괴하고 날 선 표정으로 서 있는 전경.\n\nLOCATION (lock): Before the blood-smeared boulder in the forest ritual clearing, lit by flickering campfires. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The ritual rock remains bloodstained, Youngran remains in her red shaman’s robe, and Jihu’s pendant remains on the altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 일렁이는 모닥불 불빛과 핏빛 바위 앞에서 강영란(흰 머리 여자)이 기괴하고 날 선 표정으로 서 있는 전경.\n\nLOCATION (lock): Before the blood-smeared boulder in the forest ritual clearing, lit by flickering campfires. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The ritual rock remains bloodstained, Youngran remains in her red shaman’s robe, and Jihu’s pendant remains on the altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S78sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S78sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L27B04.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지문에 언급되지 않은 인물이 화면 우측 전경에 추가되어 심각한 규칙 위반입니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 장소에서 흰 머리와 붉은 무복을 입은 인물의 기괴한 표정과 전경 구도를 규칙 위반 없이 정확하게 재현했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지문에 언급되지 않은 인물이 화면 우측 전경에 추가되어 심각한 규칙 위반입니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 장소에서 흰 머리와 붉은 무복을 입은 인물의 기괴한 표정과 전경 구도를 규칙 위반 없이 정확하게 재현했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시된 인물과 배경, 분위기를 정확히 구현했으며 불필요한 추가 인물 없이 프롬프트를 충실히 따랐습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "프롬프트에 지시되지 않은 인물이 화면 우측 전경에 등장하여 지시사항을 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시된 인물과 배경, 분위기를 정확히 구현했으며 불필요한 추가 인물 없이 프롬프트를 충실히 따랐습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "프롬프트에 지시되지 않은 인물이 화면 우측 전경에 등장하여 지시사항을 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "지문에 언급되지 않은 인물이 화면 우측 전경에 추가되어 심각한 규칙 위반입니다."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지정된 장소에서 흰 머리와 붉은 무복을 입은 인물의 기괴한 표정과 전경 구도를 규칙 위반 없이 정확하게 재현했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L27B04.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "제단(돌 탁자) 위에 유지되어야 하는 지후의 펜던트가 보이지 않습니다.",
     "fix_en": "Add Jihu's pendant resting on the surface of the stone altar table behind the woman.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add Jihu's pendant resting on the surface of the stone altar table behind the woman.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S78sh1__bgfirst_bg.png",
   "bg_asset_id": "34feb538-20dc-4dd5-81eb-31cb517203ed",
   "bg_record_key": "S78sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S78sh4": {
  "input_fingerprint": "2211a7016f6869aa",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 굳은 결심을 한 고산들이 강영란(흰 머리 여자)을 향해 고개를 깊이 숙인 자세.\n\nLOCATION (lock): In the ritual clearing facing the officiant before the blood-marked boulder. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the firelit boulder, ritual clearing, smoke, and nighttime forest. Exclude the elder's solitary command pose and show the man bowing before her.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains on the ritual altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 6,
   "C": 5
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "남성이 깊이 고개 숙인 자세를 자연스럽게 구현했으며, 기준 이미지의 세팅을 정확히 유지했습니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "세팅 유지는 양호하나, 남성의 팔이 툭 떨어져 늘어진 인사 자세가 다소 부자연스럽습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "배경 연속성은 좋으나, 인사가 깊지 않고 지문에 없는 합장 자세가 추가되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S78sh1_sel.png"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S78sh1"
  }
 },
 "S78sh9": {
  "input_fingerprint": "914d2e3210db7a62",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 예상치 못한 강영란(흰 머리 여자)의 말에 미간을 구기며 얼굴이 차갑게 굳어버린 고산들의 클로즈업.\n\nLOCATION (lock): At the edge of the firelit boulder altar in the nighttime forest clearing. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the ritual fires, blood-marked boulder, and the two figures' positions. Exclude the man's bowed posture and focus on his hardened reaction.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains on the ritual altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 4,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지시된 클로즈업 프레이밍과 차갑게 굳은 표정을 훌륭히 소화했으며, 사당과 제단의 공간적 배치 및 제단 위 펜던트까지 레퍼런스와 일관되게 유지했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "카메라가 여성의 시점으로 이동했음에도 사당과 제단이 인물 뒤에 위치하는 공간적 오류가 발생했고, 펜던트의 형태도 불명확합니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "클로즈업이 아닌 미디엄 샷에 가깝게 연출되었으며, A와 마찬가지로 세트의 공간적 논리를 어기고 사당과 제단을 인물 뒤로 잘못 배치했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S78sh4_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "레퍼런스에서 펜던트는 제단 탁자 위 중앙에 평평하게 놓여 있으나, 현재 이미지에서는 탁자 우측 모서리에 수직으로 매달려 있어 상태가 일치하지 않습니다.",
     "fix_en": "Change the gold pendant on the right edge of the background table so that it lies flat on the top surface instead of hanging vertically over the edge.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "공간 연속성 오류: 레퍼런스에서 화로는 남자 앞에 위치하지만, 현재 시점에서는 남자의 등 뒤(어깨 바로 뒤)에 배치되어 있습니다.",
     "fix_en": "Remove the fire bowl and its flames from behind the man's left shoulder.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the gold pendant on the right edge of the background table so that it lies flat on the top surface instead of hanging vertically over the edge.\n- Remove the fire bowl and its flames from behind the man's left shoulder.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S78sh4"
  }
 },
 "S79sh1::bgfirst_bg": {
  "input_fingerprint": "c4164cd96168787f",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 밝은 조명이 켜진 선착장 어선 주변으로 금월도 마을 청년들이 빽빽하게 모여든 전경.\n\nLOCATION (lock): On the brightly lit island pier around a moored fishing boat at night.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the elevated wide position, the opening crane frame looks down across the brightly illuminated dock toward the moored fishing boat and the dense gathering around it. The boat lies across the middle of the image while clusters of 금월도 마을 청년 pack the surrounding dock, establishing both the search scale and the route the camera will later descend into.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: dense gathering of village youths in the middle-center of the frame, midground; moored fishing boat in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 황 씨의 낚시 어선 (Moored) — Its broad side and deck are visible from the elevated camera position; used as The moored vessel forms the central organizing mass around which the youths gather; 선착장 (Brightly illuminated); used as The dock defines the gathering area and the route from which the search parties will depart; 선착장 조명 (On and bright); used as The dock lighting makes the crowd distribution and vessel clearly readable from above.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 밝은 조명이 켜진 선착장 어선 주변으로 금월도 마을 청년들이 빽빽하게 모여든 전경.\n\nLOCATION (lock): On the brightly lit island pier around a moored fishing boat at night.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the elevated wide position, the opening crane frame looks down across the brightly illuminated dock toward the moored fishing boat and the dense gathering around it. The boat lies across the middle of the image while clusters of 금월도 마을 청년 pack the surrounding dock, establishing both the search scale and the route the camera will later descend into.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: dense gathering of village youths in the middle-center of the frame, midground; moored fishing boat in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 황 씨의 낚시 어선 (Moored) — Its broad side and deck are visible from the elevated camera position; used as The moored vessel forms the central organizing mass around which the youths gather; 선착장 (Brightly illuminated); used as The dock defines the gathering area and the route from which the search parties will depart; 선착장 조명 (On and bright); used as The dock lighting makes the crowd distribution and vessel clearly readable from above.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S79sh1__bgfirst_bg.png",
  "asset_id": "10aceaf7-ce7f-4595-ba7d-9707eca619ce",
  "input_asset_ids": [
   "baf707fb-57fe-42fb-852c-24b0e240e642",
   "dda47e7c-bf17-4f91-8e97-7bf6000f938b"
  ]
 },
 "S79sh1": {
  "input_fingerprint": "c6999e420564c9ab",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 밝은 조명이 켜진 선착장 어선 주변으로 금월도 마을 청년들이 빽빽하게 모여든 전경.\n\nLOCATION (lock): On the brightly lit island pier around a moored fishing boat at night. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Hwang’s fishing boat remains secured at the illuminated pier while the search parties guard it. Jihu, Inwoo, and Minsuk remain coated in goat blood.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 밝은 조명이 켜진 선착장 어선 주변으로 금월도 마을 청년들이 빽빽하게 모여든 전경.\n\nLOCATION (lock): On the brightly lit island pier around a moored fishing boat at night. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Hwang’s fishing boat remains secured at the illuminated pier while the search parties guard it. Jihu, Inwoo, and Minsuk remain coated in goat blood.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 밝은 조명이 켜진 선착장 어선 주변으로 금월도 마을 청년들이 빽빽하게 모여든 전경.\n\nLOCATION (lock): On the brightly lit island pier around a moored fishing boat at night. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Hwang’s fishing boat remains secured at the illuminated pier while the search parties guard it. Jihu, Inwoo, and Minsuk remain coated in goat blood.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S79sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S79sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_ritual_clearing_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 레퍼런스 이미지의 고인돌/바위 구조물을 선착장에 정확히 반영했으며, 피 묻은 인물들과 빽빽하게 모여든 군중의 모습(전경)을 충실히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "레퍼런스 이미지의 필수 구조물이 완전히 누락되었으며, 군중이 빽빽하게 모여있지 않고 산발적으로 서 있어 프롬프트 지시를 어겼습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 레퍼런스 이미지의 고인돌/바위 구조물을 선착장에 정확히 반영했으며, 피 묻은 인물들과 빽빽하게 모여든 군중의 모습(전경)을 충실히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "레퍼런스 이미지의 필수 구조물이 완전히 누락되었으며, 군중이 빽빽하게 모여있지 않고 산발적으로 서 있어 프롬프트 지시를 어겼습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "레퍼런스 이미지의 바위와 고인돌 구조물을 선착장에 정확히 결합하였고, 피 묻은 인물들과 밀집된 군중의 야간 전경을 훌륭하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "지정된 로케이션 레퍼런스의 바위 구조물이 장소에 전혀 반영되지 않아 필수 지침을 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "레퍼런스 이미지의 바위와 고인돌 구조물을 선착장에 정확히 결합하였고, 피 묻은 인물들과 밀집된 군중의 야간 전경을 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지정된 로케이션 레퍼런스의 바위 구조물이 장소에 전혀 반영되지 않아 필수 지침을 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "지정된 레퍼런스 이미지의 고인돌/바위 구조물을 선착장에 정확히 반영했으며, 피 묻은 인물들과 빽빽하게 모여든 군중의 모습(전경)을 충실히 구현했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "레퍼런스 이미지의 필수 구조물이 완전히 누락되었으며, 군중이 빽빽하게 모여있지 않고 산발적으로 서 있어 프롬프트 지시를 어겼습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_ritual_clearing_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "어선 좌측 선체에 텍스트가 적혀 있어 '텍스트 없음' 지시를 위반했습니다.",
     "fix_en": "Remove the text and numbers from the left hull of the fishing boat.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the text and numbers from the left hull of the fishing boat.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S79sh1__bgfirst_bg.png",
   "bg_asset_id": "10aceaf7-ce7f-4595-ba7d-9707eca619ce",
   "bg_record_key": "S79sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S79sh3": {
  "input_fingerprint": "91f64f3b07d35c57",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 금월도 마을 주민들이 플래시 불빛을 어두운 바위 틈새 한 곳으로 고정해 비춘 채 서 있는 찰나.\n\nLOCATION (lock): Among the dark village lanes and rocky edges branching away from the illuminated pier. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the brightly lit pier, moored fishing boat, gathered residents, and deep night. Exclude the broad boat-centered grouping and show searchlights fixed on a rocky gap.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The villagers continue carrying flashlights and lamps during the search; the three fugitives remain masked in goat blood.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "레퍼런스의 장소 구조를 유지하며 바위 틈새를 비추는 주민들의 핵심 지문을 정확히 연출했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "레퍼런스에 전혀 존재하지 않는 건물을 화면 왼쪽에 임의로 생성하여 장소 일관성을 위반했습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "부두 바닥과 바위에 레퍼런스에 없던 거대한 균열을 만들어 고정된 지형 구조를 심각하게 훼손했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S79sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "오른쪽 바위 앞의 스탠드 조명이 크고 둥근 형태인데, 레퍼런스 이미지에 설정된 삼각대 위의 직사각형 듀얼 조명 형태와 일치하지 않습니다.",
     "fix_en": "Replace the large round lamp on the right with a double rectangular lamp mounted on a tripod to match the reference.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "가운데 검은색 재킷을 입은 남자가 쥐고 있는 손전등과 손가락의 형태가 뭉개져 기기와 융합되어 있습니다.",
     "fix_en": "Redraw the center man's hand to show distinct fingers wrapping naturally around the cylindrical body of the flashlight.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "피 묻은 재킷을 입은 남자가 들고 있는 손전등의 몸체 방향과 뻗어나가는 빛줄기의 방향이 일치하지 않고 어긋나 있습니다.",
     "fix_en": "Align the flashlight body in the blood-stained jacket man's hand so the light beam emits directly straight from its front lens.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the large round lamp on the right with a double rectangular lamp mounted on a tripod to match the reference.\n- Redraw the center man's hand to show distinct fingers wrapping naturally around the cylindrical body of the flashlight.\n- Align the flashlight body in the blood-stained jacket man's hand so the light beam emits directly straight from its front lens.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S79sh1"
  }
 },
 "S80sh14::bgfirst_bg": {
  "input_fingerprint": "9ea5c63d005e1aeb",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 창고 구석의 낡은 그물 더미 틈새로 피범벅이 된 강민숙(홍은미)의 굳은 얼굴이 슬며시 드러난 순간.\n\nLOCATION (lock): In a cramped corner of the dark fishing-gear warehouse, concealed behind stacked old nets.\n\nTIME OF DAY (lock): night, with historical day-and-night montage.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the final chest-height position beside the concealed group, the lens tightens through a narrow opening in the old net pile, with blurred net edges occupying only the side margins. Kang Min-sook's blood-covered face sits near center in a close, eye-level composition, her crouched body largely concealed while the aisle and closed door remain faintly legible behind her.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 강민숙(홍은미) in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 낡은 그물 더미 (piled in the rear corner); used as The narrow opening provides a through-object frame around Min-sook's face; 어구 창고 통로와 쌓인 어구 장비 (densely stacked) — The aisle recedes from the concealed corner toward the doorway behind the camera axis; used as Maintains spatial continuity with the completed search route behind the reveal; 창고 문 (closed) — Its inner face is seen at the far end of the aisle; used as A distant background endpoint confirming that the searcher has left.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 창고 구석의 낡은 그물 더미 틈새로 피범벅이 된 강민숙(홍은미)의 굳은 얼굴이 슬며시 드러난 순간.\n\nLOCATION (lock): In a cramped corner of the dark fishing-gear warehouse, concealed behind stacked old nets.\n\nTIME OF DAY (lock): night, with historical day-and-night montage.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the final chest-height position beside the concealed group, the lens tightens through a narrow opening in the old net pile, with blurred net edges occupying only the side margins. Kang Min-sook's blood-covered face sits near center in a close, eye-level composition, her crouched body largely concealed while the aisle and closed door remain faintly legible behind her.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 강민숙(홍은미) in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 낡은 그물 더미 (piled in the rear corner); used as The narrow opening provides a through-object frame around Min-sook's face; 어구 창고 통로와 쌓인 어구 장비 (densely stacked) — The aisle recedes from the concealed corner toward the doorway behind the camera axis; used as Maintains spatial continuity with the completed search route behind the reveal; 창고 문 (closed) — Its inner face is seen at the far end of the aisle; used as A distant background endpoint confirming that the searcher has left.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S80sh14__bgfirst_bg.png",
  "asset_id": "fe978ba5-5df8-416a-8a52-9bbc1bd6ad5c",
  "input_asset_ids": [
   "434bb50f-6350-4eec-9a50-1b0f2fe42681",
   "5920c8a3-9754-40bd-96db-d86a0391e5df"
  ]
 },
 "S80sh14": {
  "input_fingerprint": "695403d411c771ad",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, with historical day-and-night montage.\n\nSHOT TEXT (authoritative, Korean): 어두운 창고 구석의 낡은 그물 더미 틈새로 피범벅이 된 강민숙(홍은미)의 굳은 얼굴이 슬며시 드러난 순간.\n\nLOCATION (lock): In a cramped corner of the dark fishing-gear warehouse, concealed behind stacked old nets. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain covered in goat blood while hiding among the nets; Inwoo’s injured arm remains cast and Jihu’s pendant remains at the altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, with historical day-and-night montage.\n\nSHOT TEXT (authoritative, Korean): 어두운 창고 구석의 낡은 그물 더미 틈새로 피범벅이 된 강민숙(홍은미)의 굳은 얼굴이 슬며시 드러난 순간.\n\nLOCATION (lock): In a cramped corner of the dark fishing-gear warehouse, concealed behind stacked old nets. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain covered in goat blood while hiding among the nets; Inwoo’s injured arm remains cast and Jihu’s pendant remains at the altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, with historical day-and-night montage.\n\nSHOT TEXT (authoritative, Korean): 어두운 창고 구석의 낡은 그물 더미 틈새로 피범벅이 된 강민숙(홍은미)의 굳은 얼굴이 슬며시 드러난 순간.\n\nLOCATION (lock): In a cramped corner of the dark fishing-gear warehouse, concealed behind stacked old nets. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain covered in goat blood while hiding among the nets; Inwoo’s injured arm remains cast and Jihu’s pendant remains at the altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S80sh14__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S80sh14.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_dock_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "그물 더미 틈새로 피범벅이 된 굳은 얼굴이 드러나는 숏의 밀착된 구도를 지문 그대로 매우 정확하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "그물 더미 틈새가 아니라 그물을 뒤집어쓴 채 엎드린 형태로 연출되어 요구된 프레이밍과 묘사에서 벗어났습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "그물 더미 틈새로 피범벅이 된 굳은 얼굴이 드러나는 숏의 밀착된 구도를 지문 그대로 매우 정확하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "그물 더미 틈새가 아니라 그물을 뒤집어쓴 채 엎드린 형태로 연출되어 요구된 프레이밍과 묘사에서 벗어났습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "전경의 아웃포커싱된 그물 더미 틈새로 피범벅이 된 굳은 얼굴이 드러나는 숏의 구도와 깊이감을 매우 정확하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "피범벅이 된 얼굴과 어두운 창고의 분위기는 좋으나, 그물 더미 틈새라기보다는 그물을 텐트처럼 뒤집어쓴 형태에 가까워 묘사된 구도와 다소 차이가 있습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "전경의 아웃포커싱된 그물 더미 틈새로 피범벅이 된 굳은 얼굴이 드러나는 숏의 구도와 깊이감을 매우 정확하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "피범벅이 된 얼굴과 어두운 창고의 분위기는 좋으나, 그물 더미 틈새라기보다는 그물을 텐트처럼 뒤집어쓴 형태에 가까워 묘사된 구도와 다소 차이가 있습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 10
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 10
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "그물 더미 틈새로 피범벅이 된 굳은 얼굴이 드러나는 숏의 밀착된 구도를 지문 그대로 매우 정확하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "그물 더미 틈새가 아니라 그물을 뒤집어쓴 채 엎드린 형태로 연출되어 요구된 프레이밍과 묘사에서 벗어났습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_dock_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "기준 배경 이미지에서 초점이 나간 상태로 어둡게 실루엣만 보이던 전경의 좌우 형태 위에 선명한 그물 텍스처가 임의로 덧그려져 배경이 변형되었습니다.",
     "fix_en": "Restore the exact out-of-focus, dark silhouettes of the foreground masses on the extreme left and right from the reference background, removing the sharp net textures newly drawn over them.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore the exact out-of-focus, dark silhouettes of the foreground masses on the extreme left and right from the reference background, removing the sharp net textures newly drawn over them.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S80sh14__bgfirst_bg.png",
   "bg_asset_id": "fe978ba5-5df8-416a-8a52-9bbc1bd6ad5c",
   "bg_record_key": "S80sh14::bgfirst_bg",
   "chain_winner": true,
   "authority": "seed_bg"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "groupbg::기근기_마당": {
  "input_fingerprint": "539cf0244c1e1c2e",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "기근기_마당",
    "tags": [
     "S80sh31"
    ]
   },
   "context_sig": "6b01f80d76ff7278"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: In the dirt yard of a famine-era thatched home beneath a bright full moon.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 초가집 마당(밤): 보름달이 환하게 뜬 밤, 낮에 남편 시체를 먹었던 아낙이 온몸을 비튼다.\n\nTIME OF DAY (lock): night, with historical day-and-night montage.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: In the dirt yard of a famine-era thatched home beneath a bright full moon.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 초가집 마당(밤): 보름달이 환하게 뜬 밤, 낮에 남편 시체를 먹었던 아낙이 온몸을 비튼다.\n\nTIME OF DAY (lock): night, with historical day-and-night montage.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_기근기_마당_9a619f.png",
  "asset_id": "eeea1df8-8fae-4cc3-bbc9-8b734cbf01c6",
  "input_asset_ids": [
   "99463c29-4560-41ad-8b10-37d272f2f3df"
  ],
  "origin_tag": "S80sh31",
  "place_text": "In the dirt yard of a famine-era thatched home beneath a bright full moon.",
  "origin_inputs": {
   "place_text": "In the dirt yard of a famine-era thatched home beneath a bright full moon.",
   "time_of_day_en": "night, with historical day-and-night montage",
   "conti_asset_id": "99463c29-4560-41ad-8b10-37d272f2f3df"
  }
 },
 "S80sh31::bgfirst_bg": {
  "input_fingerprint": "b4d75bdff4178d27",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 고통에 찬 비명을 지르듯 입을 크게 벌리고, 두 눈이 새빨갛게 충혈된 경신 대기근의 아낙의 끔찍한 얼굴 클로즈업.\n\nLOCATION (lock): In the dirt yard of a famine-era thatched home beneath a bright full moon.\n\nTIME OF DAY (lock): night, with historical day-and-night montage.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: The dolly-in completes at an extreme facial distance after rising from the woman's kneeling level, holding her open mouth and bloodshot eyes near the center while a trace of her contorted shoulders remains along the lower edge. The almost frontal eye-level endpoint turns the transformation into a confrontational image ready for the hard cut.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 경신 대기근의 아낙 in the middle-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 초가집 마당 (visible only at the frame edges); used as A narrow peripheral indication of the location behind the transformed face; 보름달 (bright above the courtyard); used as Provides the explicitly described night illumination while remaining outside or near the upper frame boundary.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 고통에 찬 비명을 지르듯 입을 크게 벌리고, 두 눈이 새빨갛게 충혈된 경신 대기근의 아낙의 끔찍한 얼굴 클로즈업.\n\nLOCATION (lock): In the dirt yard of a famine-era thatched home beneath a bright full moon.\n\nTIME OF DAY (lock): night, with historical day-and-night montage.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: The dolly-in completes at an extreme facial distance after rising from the woman's kneeling level, holding her open mouth and bloodshot eyes near the center while a trace of her contorted shoulders remains along the lower edge. The almost frontal eye-level endpoint turns the transformation into a confrontational image ready for the hard cut.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 경신 대기근의 아낙 in the middle-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 초가집 마당 (visible only at the frame edges); used as A narrow peripheral indication of the location behind the transformed face; 보름달 (bright above the courtyard); used as Provides the explicitly described night illumination while remaining outside or near the upper frame boundary.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S80sh31__bgfirst_bg.png",
  "asset_id": "a29f4218-c9da-42ac-81a8-7214b18ff734",
  "input_asset_ids": [
   "99463c29-4560-41ad-8b10-37d272f2f3df",
   "eeea1df8-8fae-4cc3-bbc9-8b734cbf01c6"
  ]
 },
 "S80sh31": {
  "input_fingerprint": "fd48d882d9195d87",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, with historical day-and-night montage.\n\nSHOT TEXT (authoritative, Korean): 고통에 찬 비명을 지르듯 입을 크게 벌리고, 두 눈이 새빨갛게 충혈된 경신 대기근의 아낙의 끔찍한 얼굴 클로즈업.\n\nLOCATION (lock): In the dirt yard of a famine-era thatched home beneath a bright full moon. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, with historical day-and-night montage.\n\nSHOT TEXT (authoritative, Korean): 고통에 찬 비명을 지르듯 입을 크게 벌리고, 두 눈이 새빨갛게 충혈된 경신 대기근의 아낙의 끔찍한 얼굴 클로즈업.\n\nLOCATION (lock): In the dirt yard of a famine-era thatched home beneath a bright full moon. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, with historical day-and-night montage.\n\nSHOT TEXT (authoritative, Korean): 고통에 찬 비명을 지르듯 입을 크게 벌리고, 두 눈이 새빨갛게 충혈된 경신 대기근의 아낙의 끔찍한 얼굴 클로즈업.\n\nLOCATION (lock): In the dirt yard of a famine-era thatched home beneath a bright full moon. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S80sh31__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S80sh31.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_기근기_마당_9a619f.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 로케이션 내에서 사실적인 카메라 구도를 잡았으며, 과장 없는 자연스러운 충혈된 눈과 비명 짓는 표정 연기를 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "클로즈업과 충혈된 눈의 조건은 맞췄으나, 안구가 비현실적으로 과장되어 '자연스러운 인간의 연기'라는 제약 사항을 위반했으며 배경이 평면적으로 합성된 느낌을 줍니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 로케이션 내에서 사실적인 카메라 구도를 잡았으며, 과장 없는 자연스러운 충혈된 눈과 비명 짓는 표정 연기를 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "클로즈업과 충혈된 눈의 조건은 맞췄으나, 안구가 비현실적으로 과장되어 '자연스러운 인간의 연기'라는 제약 사항을 위반했으며 배경이 평면적으로 합성된 느낌을 줍니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시문에 따른 클로즈업 프레이밍과 충혈된 눈, 고통스럽게 절규하는 표정을 해부학적 오류 없이 사실적으로 잘 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "아래턱에 치아가 비정상적으로 많이 생성되는 치명적인 해부학적 오류(Hard Violation)가 발생하여 실사 사진으로서의 사실성이 떨어집니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시문에 따른 클로즈업 프레이밍과 충혈된 눈, 고통스럽게 절규하는 표정을 해부학적 오류 없이 사실적으로 잘 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "아래턱에 치아가 비정상적으로 많이 생성되는 치명적인 해부학적 오류(Hard Violation)가 발생하여 실사 사진으로서의 사실성이 떨어집니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 7,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 7,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "지정된 로케이션 내에서 사실적인 카메라 구도를 잡았으며, 과장 없는 자연스러운 충혈된 눈과 비명 짓는 표정 연기를 훌륭하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "클로즈업과 충혈된 눈의 조건은 맞췄으나, 안구가 비현실적으로 과장되어 '자연스러운 인간의 연기'라는 제약 사항을 위반했으며 배경이 평면적으로 합성된 느낌을 줍니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_기근기_마당_9a619f.png"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S80sh31__bgfirst_bg.png",
   "bg_asset_id": "a29f4218-c9da-42ac-81a8-7214b18ff734",
   "bg_record_key": "S80sh31::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "기근기_마당",
   "groupbg_asset_id": "eeea1df8-8fae-4cc3-bbc9-8b734cbf01c6"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "groupbg::과거_상륙해변": {
  "input_fingerprint": "0e8cbbe3c70330e4",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "과거_상륙해변",
    "tags": [
     "S80sh56"
    ]
   },
   "context_sig": "4e78a29f8c2b1531"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On a sandy island shore at the landing point, where arriving survivors face the waiting residents.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 금월도 입구(낮): 배를 타고 들어오는 인간과 똑같은 모습의 십수 명 육혈귀들.\n- 바닷가에 나와 이들을 반갑게 맞는 금월도 원주민들의 해맑은 모습.\n\nTIME OF DAY (lock): night, with historical day-and-night montage.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On a sandy island shore at the landing point, where arriving survivors face the waiting residents.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 금월도 입구(낮): 배를 타고 들어오는 인간과 똑같은 모습의 십수 명 육혈귀들.\n- 바닷가에 나와 이들을 반갑게 맞는 금월도 원주민들의 해맑은 모습.\n\nTIME OF DAY (lock): night, with historical day-and-night montage.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_과거_상륙해변_faa2c8.png",
  "asset_id": "5b6f50e7-8a63-4e2b-a37c-4a60bd5386c2",
  "input_asset_ids": [
   "f99f8466-40c1-4bbb-a8c7-6b1415f03157"
  ],
  "origin_tag": "S80sh56",
  "place_text": "On a sandy island shore at the landing point, where arriving survivors face the waiting residents.",
  "origin_inputs": {
   "place_text": "On a sandy island shore at the landing point, where arriving survivors face the waiting residents.",
   "time_of_day_en": "night, with historical day-and-night montage",
   "conti_asset_id": "f99f8466-40c1-4bbb-a8c7-6b1415f03157"
  }
 },
 "S80sh56::bgfirst_bg": {
  "input_fingerprint": "443fc3c1aa031f90",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 해변 모래사장에 한쪽 발을 내디딘 자세로 원주민들을 향해 섬뜩한 미소를 머금은 과거의 조선인 육혈귀 생존자들의 얼굴.\n\nLOCATION (lock): On a sandy island shore at the landing point, where arriving survivors face the waiting residents.\n\nTIME OF DAY (lock): night, with historical day-and-night montage.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At the end of the backward shoreline track, the camera settles slightly below eye level in a wide composition that favors the survivors' front rank without losing their leading feet planting on the sand. Their outwardly human faces occupy the upper middle of the frame as they smile past the lens toward the waiting islanders, who remain farther inland as a welcoming counter-group.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 과거의 조선인 육혈귀 생존자 in the middle-center of the frame, midground, moves toward waiting islanders; 금월도 마을 주민 in the upper-center of the frame, background; shoreline landing area in the lower-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 해변 모래사장 (visible beneath the arriving survivors); used as The lower-frame landing surface preserves the visual evidence of the first step onto the island; 배 (arrived at the island entrance) — The side nearest the shore is visible behind the disembarking group; used as A scale and arrival reference behind the survivors; 바다 (visible in daylight); used as Separates the boat from the island shore across the rear of the composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 해변 모래사장에 한쪽 발을 내디딘 자세로 원주민들을 향해 섬뜩한 미소를 머금은 과거의 조선인 육혈귀 생존자들의 얼굴.\n\nLOCATION (lock): On a sandy island shore at the landing point, where arriving survivors face the waiting residents.\n\nTIME OF DAY (lock): night, with historical day-and-night montage.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At the end of the backward shoreline track, the camera settles slightly below eye level in a wide composition that favors the survivors' front rank without losing their leading feet planting on the sand. Their outwardly human faces occupy the upper middle of the frame as they smile past the lens toward the waiting islanders, who remain farther inland as a welcoming counter-group.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 과거의 조선인 육혈귀 생존자 in the middle-center of the frame, midground, moves toward waiting islanders; 금월도 마을 주민 in the upper-center of the frame, background; shoreline landing area in the lower-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 해변 모래사장 (visible beneath the arriving survivors); used as The lower-frame landing surface preserves the visual evidence of the first step onto the island; 배 (arrived at the island entrance) — The side nearest the shore is visible behind the disembarking group; used as A scale and arrival reference behind the survivors; 바다 (visible in daylight); used as Separates the boat from the island shore across the rear of the composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S80sh56__bgfirst_bg.png",
  "asset_id": "97297b1e-e024-4552-8bf1-adb7e356d4ab",
  "input_asset_ids": [
   "f99f8466-40c1-4bbb-a8c7-6b1415f03157",
   "5b6f50e7-8a63-4e2b-a37c-4a60bd5386c2"
  ]
 },
 "S80sh56": {
  "input_fingerprint": "1eeeb68017922051",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, with historical day-and-night montage.\n\nSHOT TEXT (authoritative, Korean): 해변 모래사장에 한쪽 발을 내디딘 자세로 원주민들을 향해 섬뜩한 미소를 머금은 과거의 조선인 육혈귀 생존자들의 얼굴.\n\nLOCATION (lock): On a sandy island shore at the landing point, where arriving survivors face the waiting residents. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, with historical day-and-night montage.\n\nSHOT TEXT (authoritative, Korean): 해변 모래사장에 한쪽 발을 내디딘 자세로 원주민들을 향해 섬뜩한 미소를 머금은 과거의 조선인 육혈귀 생존자들의 얼굴.\n\nLOCATION (lock): On a sandy island shore at the landing point, where arriving survivors face the waiting residents. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, with historical day-and-night montage.\n\nSHOT TEXT (authoritative, Korean): 해변 모래사장에 한쪽 발을 내디딘 자세로 원주민들을 향해 섬뜩한 미소를 머금은 과거의 조선인 육혈귀 생존자들의 얼굴.\n\nLOCATION (lock): On a sandy island shore at the landing point, where arriving survivors face the waiting residents. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S80sh56__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S80sh56.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_과거_상륙해변_faa2c8.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "한쪽 발을 모래사장에 내디딘 자세와 섬뜩한 미소, 원주민을 마주 보는 공간적 구도를 지시문대로 정확하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "원주민을 향하는 대신 카메라를 향해 걸어오고 있으며, 샷이 너무 넓어 요구된 표정이 잘 드러나지 않습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "한쪽 발을 모래사장에 내디딘 자세와 섬뜩한 미소, 원주민을 마주 보는 공간적 구도를 지시문대로 정확하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "원주민을 향하는 대신 카메라를 향해 걸어오고 있으며, 샷이 너무 넓어 요구된 표정이 잘 드러나지 않습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지정된 현대식 항구 로케이션을 무시하고 초가집 해변으로 변경했으며, 주인공 다리 뒤에 허공에 뜬 여분의 다리가 생성되는 치명적인 신체 구조 오류(Hard Violation)가 발생했습니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "레퍼런스의 콘크리트 슬립웨이와 항구 배경을 완벽하게 유지했으나, '얼굴'에 집중해야 하는 프레이밍을 무시하고 피사체가 작게 보이는 와이드 풀샷으로 렌더링하여 샷 크기 지시를 놓쳤습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "지정된 현대식 항구 로케이션을 무시하고 초가집 해변으로 변경했으며, 주인공 다리 뒤에 허공에 뜬 여분의 다리가 생성되는 치명적인 신체 구조 오류(Hard Violation)가 발생했습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "레퍼런스의 콘크리트 슬립웨이와 항구 배경을 완벽하게 유지했으나, '얼굴'에 집중해야 하는 프레이밍을 무시하고 피사체가 작게 보이는 와이드 풀샷으로 렌더링하여 샷 크기 지시를 놓쳤습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 9,
     "B": 10
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 9,
   "B": 10
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "한쪽 발을 모래사장에 내디딘 자세와 섬뜩한 미소, 원주민을 마주 보는 공간적 구도를 지시문대로 정확하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "원주민을 향하는 대신 카메라를 향해 걸어오고 있으며, 샷이 너무 넓어 요구된 표정이 잘 드러나지 않습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_과거_상륙해변_faa2c8.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "레퍼런스 사진의 콘크리트 부두와 현대식 구조물 대신 모래사장과 전통 초가집이 렌더링되어 장소의 물리적 환경이 일치하지 않습니다.",
     "fix_en": "Replace the sandy beach and traditional hut with the concrete ground and modern harbor elements exactly as shown in the location reference.",
     "needs_regeneration": true,
     "unfixable": true
    },
    {
     "issue_ko": "지문은 '원주민들을 향해' 미소를 지으라고 했으나, 왼쪽의 메인 캐릭터가 카메라 렌즈를 정면으로 응시하고 있습니다.",
     "fix_en": "Redirect the main character's head and gaze to look at the group of natives on the right instead of the camera.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "왼쪽 보트에서 내리는 메인 캐릭터의 두 다리 사이에 정체불명의 세 번째 다리(맨발)가 있습니다.",
     "fix_en": "Remove the extra bare leg and foot visible between the legs of the main character on the left.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redirect the main character's head and gaze to look at the group of natives on the right instead of the camera.\n- Remove the extra bare leg and foot visible between the legs of the main character on the left.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S80sh56__bgfirst_bg.png",
   "bg_asset_id": "97297b1e-e024-4552-8bf1-adb7e356d4ab",
   "bg_record_key": "S80sh56::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "과거_상륙해변",
   "groupbg_asset_id": "5b6f50e7-8a63-4e2b-a37c-4a60bd5386c2"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S81sh1": {
  "input_fingerprint": "575091a9b69e8eb6",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)와 정인우가 강민숙(홍은미)을 향해 경악으로 입을 떡 벌린 얼굴.\n\nLOCATION (lock): Inside the dark fishing-gear warehouse, hidden among nets, rods, traps, and stacked baskets. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain coated in goat blood inside the gear shed; Jihu’s pendant remains at the ritual altar.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상); 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 5,
   "A": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "요구된 '얼굴' 프레이밍을 완벽히 구현했고 이전 샷의 V자 그물 틈새 위치를 정확히 유지했으나, 정인우의 모자 누락이 아쉽습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "얼굴 위주의 프레이밍은 적절하나 이전 샷의 그물 배치 연속성이 떨어지며 정인우의 모자가 누락되었습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "정인우의 복장은 정확하나, '얼굴' 샷을 너무 넓게 잡아 프레이밍을 위반했고 지후가 그물 밖으로 이동해 연속성을 깼습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S80sh14_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903809>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "배경 안쪽에 밝은 하얀 빛이 들어오는 곳이 있어, 지정된 시간대인 '밤(night)' 설정과 일치하지 않습니다.",
     "fix_en": "Change the bright white light in the background window or door to dark night shadows.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "프롬프트에 허용되지 않은 인물(또는 피가 전혀 묻지 않은 인물)이 오른쪽 앞경에 나타나, 등장인물 제한 및 '피를 뒤집어쓴' 상태 유지 규칙을 위반했습니다.",
     "fix_en": "Remove the person in the right foreground, replacing them with dark, out-of-focus fishing nets and gear to match the left foreground.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "정인우가 낚시 조끼 안에 입은 상의가 레퍼런스의 짙은 녹색 재킷이 아닌 회색 긴팔 셔츠로 잘못 묘사되었습니다.",
     "fix_en": "Change the grey long-sleeve shirt worn under Inwoo's vest to a dark green jacket.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the bright white light in the background window or door to dark night shadows.\n- Remove the person in the right foreground, replacing them with dark, out-of-focus fishing nets and gear to match the left foreground.\n- Change the grey long-sleeve shirt worn under Inwoo's vest to a dark green jacket.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S80sh14"
  }
 },
 "S81sh10": {
  "input_fingerprint": "b47be4826f0721ba",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 눈물을 가득 머금은 채 원망스러운 눈빛으로 강민숙(홍은미)을 쏘아보는 찰나.\n\nLOCATION (lock): In the concealed corner of the fishing-gear warehouse, facing the mother at close range. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the cramped fishing-gear store, stacked nets, traps, baskets, dust, and dim daylight. Exclude the shared shocked reaction and focus on the daughter's tearful glare.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The three remain covered in goat blood; Jihu’s pendant remains at the altar and Inwoo’s injured arm remains cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 9,
   "B": 6,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 9,
    "verdict_ko": "이전 샷의 의상과 배경을 정확히 유지하면서, 눈물이 맺힌 채 원망스럽게 쏘아보는 딸의 감정선을 완벽하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "지문 및 이전 샷에 없는 어깨에 손을 올리는 동작이 임의로 추가되었으며, 원망스러운 표정 연기가 다소 부족합니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "이전 샷에서 확립된 지후의 의상(어두운 겉옷과 셔츠)이 완전히 다른 옷으로 변경되어 연속성 유지에 실패했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S81sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지후가 노려보고 있는 전경의 인물(어머니)이 이전 컷의 남성이 입고 있던 황갈색 다용도 조끼와 카키색 셔츠를 그대로 입고 있습니다.",
     "fix_en": "Change the clothing of the foreground character on the left to a simple dark shirt, completely removing the tan vest and green collar that belong to the man in the previous shot.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the clothing of the foreground character on the left to a simple dark shirt, completely removing the tan vest and green collar that belong to the man in the previous shot.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S81sh1"
  }
 },
 "S81sh12": {
  "input_fingerprint": "fd393cf605a546ed",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 강민숙(홍은미)이 참지 못하고 고개를 푹 떨군 채 눈물을 뚝뚝 흘리며 얼굴을 일그러뜨린 순간.\n\nLOCATION (lock): Among the stacked fishing gear inside the dim warehouse hideout. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same crowded gear storage, dusty light, nets, and the three figures' positions. Exclude the daughter's accusatory close-up and focus on the mother's breakdown.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain blood-coated in the gear shed.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강민숙(홍은미) (한국인 40대 후반 여성, 갸름한 얼굴, 단정한 검은 머리, 귀티 나는 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 4,
   "B": 7,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "표정 묘사는 좋으나, 이전 샷에서 고정된 인물의 자세가 갑자기 낮아져(무릎을 꿇은 듯함) 물리적 연속성 규칙을 위반했습니다."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "레퍼런스의 전경 어깨걸이 구도, 조명, 배경을 완벽히 유지하면서 고개를 떨구고 눈물을 흘리는 일그러진 표정을 잘 구현했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "프레이밍이 레퍼런스의 연속성을 잃었으며, 샷 텍스트에 없는 정체불명의 인물(정장 차림)들을 화면에 추가하여 심각한 위반을 범했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S81sh10_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:844015>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "정면에서 울고 있는 인물의 얼굴과 연령대가 캐릭터 레퍼런스에 지정된 강민숙(40대 후반)이 아닌, 이전 컷의 젊은 여성으로 묘사되었습니다.",
     "fix_en": "Replace the crying woman's face and facial structure to exactly match the older woman (Kang Min-sook) from the character reference, maintaining the tears and blood.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "우는 인물의 의상이 캐릭터 레퍼런스의 강민숙 의상(카멜색 코트와 밝은 터틀넥)과 일치하지 않습니다.",
     "fix_en": "Change the dark jacket on the crying woman to the camel coat and light turtleneck shown in the character reference, ensuring the garments remain heavily coated in blood.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the crying woman's face and facial structure to exactly match the older woman (Kang Min-sook) from the character reference, maintaining the tears and blood.\n- Change the dark jacket on the crying woman to the camel coat and light turtleneck shown in the character reference, ensuring the garments remain heavily coated in blood.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S81sh10"
  }
 },
 "S82sh2": {
  "input_fingerprint": "2f27362caf6ce49a",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dawn transitioning to morning.\n\nSHOT TEXT (authoritative, Korean): 바다 위로 붉은 태양이 눈부시게 솟아오른 찰나의 전경.\n\nLOCATION (lock): Over the open sea immediately off the island, where the red sun rises through the dawn horizon. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 5,
   "C": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "제시된 장소(성산 일출봉 등)의 지형적 특징을 정확히 반영하며, 바다 위로 떠오르는 붉은 태양의 전경을 지시문대로 충실히 구현했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "붉은 태양이 떠오르는 일출의 모습은 잘 표현되었으나, 배경에 위치한 섬의 형태가 특정된 장소의 고유한 특징을 명확히 보여주지 못합니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "일출 풍경을 담고 있으나, 전체적인 화각(광각) 대비 수평선에 걸친 태양의 크기와 형태가 다소 부자연스러우며 장소의 특징도 부족합니다."
   }
  ],
  "refs": [],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S83sh4::bgfirst_bg": {
  "input_fingerprint": "fc80a4fba2a38666",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 고산들이 심각한 표정으로 강영란(흰 머리 여자)을 묵묵히 내려다보는 측면.\n\nLOCATION (lock): Inside the island restaurant's dining room, near the table and open view toward the daylight waterfront.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the seated-eye-level position beside Sandeul, the static medium frame views him in strict side profile as he looks down toward Yeongran at the table. Sandeul occupies one midground side with his weight held near the doorway, while Yeongran sits lower on the opposite side, keeping their vertical power relation legible without changing the established eyeline.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 고산들 in the middle-left of the frame, midground, looks toward Yeongran at the table; 강영란(흰 머리 여자) in the lower-right of the frame, midground; restaurant table in the lower-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 테이블 (occupied by Yeongran) — One long edge angles between Yeongran and the camera; used as Anchors Yeongran's seated position and separates her from Sandeul's doorway position; 식당 문가의 바깥 시야 (unobstructed enough to see outside) — The camera can see past Sandeul toward the sea side and searching villagers; used as Provides a view toward the villagers moving outside and locates Sandeul at the threshold.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 고산들이 심각한 표정으로 강영란(흰 머리 여자)을 묵묵히 내려다보는 측면.\n\nLOCATION (lock): Inside the island restaurant's dining room, near the table and open view toward the daylight waterfront.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the seated-eye-level position beside Sandeul, the static medium frame views him in strict side profile as he looks down toward Yeongran at the table. Sandeul occupies one midground side with his weight held near the doorway, while Yeongran sits lower on the opposite side, keeping their vertical power relation legible without changing the established eyeline.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 고산들 in the middle-left of the frame, midground, looks toward Yeongran at the table; 강영란(흰 머리 여자) in the lower-right of the frame, midground; restaurant table in the lower-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 테이블 (occupied by Yeongran) — One long edge angles between Yeongran and the camera; used as Anchors Yeongran's seated position and separates her from Sandeul's doorway position; 식당 문가의 바깥 시야 (unobstructed enough to see outside) — The camera can see past Sandeul toward the sea side and searching villagers; used as Provides a view toward the villagers moving outside and locates Sandeul at the threshold.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S83sh4__bgfirst_bg.png",
  "asset_id": "458225b1-7223-436f-8e19-62d5f8b3f646",
  "input_asset_ids": [
   "08f50216-04df-422b-93b3-2c4ea8e897af",
   "ef3e7e06-34ca-4d12-a71f-1a33e8ed21ad"
  ]
 },
 "S83sh4": {
  "input_fingerprint": "dd513b8e01958cef",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 고산들이 심각한 표정으로 강영란(흰 머리 여자)을 묵묵히 내려다보는 측면.\n\nLOCATION (lock): Inside the island restaurant's dining room, near the table and open view toward the daylight waterfront. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The shed behind the restaurant remains a burned-out ruin; Jihu’s pendant remains at the ritual altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 고산들이 심각한 표정으로 강영란(흰 머리 여자)을 묵묵히 내려다보는 측면.\n\nLOCATION (lock): Inside the island restaurant's dining room, near the table and open view toward the daylight waterfront. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The shed behind the restaurant remains a burned-out ruin; Jihu’s pendant remains at the ritual altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 고산들이 심각한 표정으로 강영란(흰 머리 여자)을 묵묵히 내려다보는 측면.\n\nLOCATION (lock): Inside the island restaurant's dining room, near the table and open view toward the daylight waterfront. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The shed behind the restaurant remains a burned-out ruin; Jihu’s pendant remains at the ritual altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S83sh4__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S83sh4.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L41B02.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 측면 구도에서 인물이 묵묵히 내려다보는 행동과 심각한 표정이 적절한 샷 크기로 명확하게 잘 표현되었습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "배경과 인물의 기본 조건은 충족했으나, 두 인물의 거리가 너무 멀어 '내려다보는 측면'의 상호작용과 심각한 표정이 효과적으로 전달되지 않습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 측면 구도에서 인물이 묵묵히 내려다보는 행동과 심각한 표정이 적절한 샷 크기로 명확하게 잘 표현되었습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "배경과 인물의 기본 조건은 충족했으나, 두 인물의 거리가 너무 멀어 '내려다보는 측면'의 상호작용과 심각한 표정이 효과적으로 전달되지 않습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "측면 구도에서 남자가 흰머리 여자를 심각하게 내려다보는 인물 간의 밀접한 상호작용을 매우 충실하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "인물 간 거리가 지나치게 멀어 '내려다보는' 행위의 몰입감이 떨어지며, 금지된 레퍼런스 이미지의 카메라 구도를 거의 그대로 모방했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "측면 구도에서 남자가 흰머리 여자를 심각하게 내려다보는 인물 간의 밀접한 상호작용을 매우 충실하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "인물 간 거리가 지나치게 멀어 '내려다보는' 행위의 몰입감이 떨어지며, 금지된 레퍼런스 이미지의 카메라 구도를 거의 그대로 모방했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 10,
     "B": 15
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 10,
   "B": 15
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "지정된 측면 구도에서 인물이 묵묵히 내려다보는 행동과 심각한 표정이 적절한 샷 크기로 명확하게 잘 표현되었습니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "배경과 인물의 기본 조건은 충족했으나, 두 인물의 거리가 너무 멀어 '내려다보는 측면'의 상호작용과 심각한 표정이 효과적으로 전달되지 않습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L41B02.png"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S83sh4__bgfirst_bg.png",
   "bg_asset_id": "458225b1-7223-436f-8e19-62d5f8b3f646",
   "bg_record_key": "S83sh4::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S83sh5": {
  "input_fingerprint": "61f4182c72534ed7",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)이 서늘하고 독기 어린 눈빛으로 허공을 뚫어지게 노려보는 얼굴 클로즈업.\n\nLOCATION (lock): At the restaurant table facing the doorway and the search activity outside in daylight. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the daylight restaurant hall, table, doorway, sea-facing view, and the two figures' positions. Exclude the man's downward look and focus on the elder's cold stare.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains under Youngran’s control at the ritual altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 4,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "요구된 얼굴 클로즈업 샷과 문밖의 수색 인력을 정확히 묘사했으며, 레퍼런스의 좌석 및 인물의 방향을 일관되게 유지했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "지정된 얼굴 클로즈업 대신 프레임이 넓어졌고, 남자의 위치가 원본과 다르게 변경되었으며 외부 수색 장면이 생략되었습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "인물이 문을 등지고 앉아 카메라를 정면으로 바라보는 구도로 렌더링되어 원본의 공간 기하학 및 인물 방향과 충돌하는 치명적인 위반(Hard Violation)입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S83sh4_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "여자의 표정이 수동적이고 약간 슬퍼 보이며, 프롬프트에서 요구한 '서늘하고 독기 어린 눈빛으로 노려보는' 강렬한 표정이 반영되지 않았습니다.",
     "fix_en": "Lower the woman's eyebrows and narrow her eyelids to create a sharp, venomous glare.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Lower the woman's eyebrows and narrow her eyelids to create a sharp, venomous glare.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S83sh4"
  }
 },
 "S84sh1::bgfirst_bg": {
  "input_fingerprint": "4ed8390a3e8c5cf5",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 먼지가 잔뜩 낀 흐릿한 창문 틈새로 바깥을 예리하게 내다보는 강민숙(홍은미)의 뒷모습.\n\nLOCATION (lock): At the dusty window inside the pier-side fishing-gear warehouse, overlooking the daylight dock.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From directly behind Min-sook at shoulder height, a static medium frame aligns her back with the dusty, obscured window as she leans forward to inspect the dock. She occupies the middle-left portion while the window spans the upper center and right, allowing a limited view of the dock through the dust without revealing the two companions behind her.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 강민숙(홍은미) in the middle-left of the frame, midground, looks toward dock beyond the dusty window; dusty window and dock beyond in the upper-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 먼지가 잔뜩 낀 흐릿한 창문 (heavily dusty and obscured) — Its dusty surface faces the camera, with the dock only indistinctly visible through it; used as The principal sightline and obscuring layer through which Min-sook monitors the exterior; 선착장 (being watched from inside) — Its visible length recedes away from the storage space beyond the window; used as The exterior surveillance target beyond the window; 쌓인 어구 장비 (densely stacked) — The stacked sides face inward toward Min-sook and the window axis; used as Frames Min-sook's silhouette from both sides while preserving the cramped hiding place.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 먼지가 잔뜩 낀 흐릿한 창문 틈새로 바깥을 예리하게 내다보는 강민숙(홍은미)의 뒷모습.\n\nLOCATION (lock): At the dusty window inside the pier-side fishing-gear warehouse, overlooking the daylight dock.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From directly behind Min-sook at shoulder height, a static medium frame aligns her back with the dusty, obscured window as she leans forward to inspect the dock. She occupies the middle-left portion while the window spans the upper center and right, allowing a limited view of the dock through the dust without revealing the two companions behind her.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 강민숙(홍은미) in the middle-left of the frame, midground, looks toward dock beyond the dusty window; dusty window and dock beyond in the upper-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 먼지가 잔뜩 낀 흐릿한 창문 (heavily dusty and obscured) — Its dusty surface faces the camera, with the dock only indistinctly visible through it; used as The principal sightline and obscuring layer through which Min-sook monitors the exterior; 선착장 (being watched from inside) — Its visible length recedes away from the storage space beyond the window; used as The exterior surveillance target beyond the window; 쌓인 어구 장비 (densely stacked) — The stacked sides face inward toward Min-sook and the window axis; used as Frames Min-sook's silhouette from both sides while preserving the cramped hiding place.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S84sh1__bgfirst_bg.png",
  "asset_id": "eb472c59-0bb5-45da-9059-fa54830abf1d",
  "input_asset_ids": [
   "622e4aec-55af-4f01-a0bd-424ed4f310c3",
   "ea2145e9-3148-47b4-b459-c45c8c0b49cc"
  ]
 },
 "S84sh1": {
  "input_fingerprint": "1eed4fec7ea5ff9d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 먼지가 잔뜩 낀 흐릿한 창문 틈새로 바깥을 예리하게 내다보는 강민숙(홍은미)의 뒷모습.\n\nLOCATION (lock): At the dusty window inside the pier-side fishing-gear warehouse, overlooking the daylight dock. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain stained with goat blood while hiding; Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 먼지가 잔뜩 낀 흐릿한 창문 틈새로 바깥을 예리하게 내다보는 강민숙(홍은미)의 뒷모습.\n\nLOCATION (lock): At the dusty window inside the pier-side fishing-gear warehouse, overlooking the daylight dock. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain stained with goat blood while hiding; Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 먼지가 잔뜩 낀 흐릿한 창문 틈새로 바깥을 예리하게 내다보는 강민숙(홍은미)의 뒷모습.\n\nLOCATION (lock): At the dusty window inside the pier-side fishing-gear warehouse, overlooking the daylight dock. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain stained with goat blood while hiding; Inwoo’s injured arm remains cast.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S84sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S84sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_open_sea_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "레퍼런스의 파란색 격자 창문 구조를 무시하고 다른 형태의 창문을 생성했으며, 피사체가 있어야 할 창고 건물을 창밖 멀리 배치하여 위치 설정을 심각하게 위반했습니다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "샷 텍스트에 명시된 인물의 뒷모습과 핏자국을 잘 포착했으며, 레퍼런스 사진의 파란색 격자 창문 형태와 먼지 낀 질감을 내부 시점에서 매우 정확하게 구현했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "레퍼런스의 파란색 격자 창문 구조를 무시하고 다른 형태의 창문을 생성했으며, 피사체가 있어야 할 창고 건물을 창밖 멀리 배치하여 위치 설정을 심각하게 위반했습니다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "샷 텍스트에 명시된 인물의 뒷모습과 핏자국을 잘 포착했으며, 레퍼런스 사진의 파란색 격자 창문 형태와 먼지 낀 질감을 내부 시점에서 매우 정확하게 구현했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "기준 사진의 파란색 다중 창틀 구조와 내부에서 부두를 내다보는 시점을 정확히 구현했으며, 지시된 뒷모습과 혈흔 설정도 잘 표현했습니다."
     },
     {
      "label": "B",
      "score": 2,
      "verdict_ko": "캐릭터가 창고 내부에 있어야 한다는 설정과 달리 외부의 다른 건물에서 창고를 바라보고 있는 심각한 로케이션 위반이며, 창문 형태도 기준 사진과 전혀 다릅니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "기준 사진의 파란색 다중 창틀 구조와 내부에서 부두를 내다보는 시점을 정확히 구현했으며, 지시된 뒷모습과 혈흔 설정도 잘 표현했습니다."
     },
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "캐릭터가 창고 내부에 있어야 한다는 설정과 달리 외부의 다른 건물에서 창고를 바라보고 있는 심각한 로케이션 위반이며, 창문 형태도 기준 사진과 전혀 다릅니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "레퍼런스의 파란색 격자 창문 구조를 무시하고 다른 형태의 창문을 생성했으며, 피사체가 있어야 할 창고 건물을 창밖 멀리 배치하여 위치 설정을 심각하게 위반했습니다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "샷 텍스트에 명시된 인물의 뒷모습과 핏자국을 잘 포착했으며, 레퍼런스 사진의 파란색 격자 창문 형태와 먼지 낀 질감을 내부 시점에서 매우 정확하게 구현했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_open_sea_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "창문 밖 선착장에 샷 텍스트에 명시되지 않은 사람들이 나타나 있습니다.",
     "fix_en": "Remove all the people standing and walking on the dock outside the window.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "창문 격자 구조가 레퍼런스 사진과 일치하지 않습니다 (레퍼런스는 4x3 대칭 구조이나 이미지는 비대칭 3열입니다).",
     "fix_en": "Reconstruct the window mullions into a symmetrical grid of 4 columns and 3 rows of panes to match the reference location.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all the people standing and walking on the dock outside the window.\n- Reconstruct the window mullions into a symmetrical grid of 4 columns and 3 rows of panes to match the reference location.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S84sh1__bgfirst_bg.png",
   "bg_asset_id": "eb472c59-0bb5-45da-9059-fa54830abf1d",
   "bg_record_key": "S84sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S84sh5": {
  "input_fingerprint": "a5f469aa3fc172e2",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 정인우가 백팩에서 38구경 권총을 불쑥 꺼내 든 찰나의 클로즈업.\n\nLOCATION (lock): Inside the hidden corner of the fishing-gear warehouse near its dusty window. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the dusty gear store, murky window, stacked nets and traps, and daylight. Exclude the mother's lookout pose and emphasize the handgun drawn from the backpack.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo now carries the .38-caliber revolver from his backpack; his injured arm remains cast, and all three remain goat-blood stained.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 6,
   "B": 8,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "배경과 인물의 상태는 기준에 부합하나, 요구된 클로즈업 대신 인물 상반신이 모두 보이는 미디엄 샷으로 연출되어 프레이밍 지시를 어겼습니다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "가방에서 총을 꺼내는 찰나의 클로즈업이라는 프레이밍 지시를 가장 정확히 구현했으며, 깁스와 혈흔 등 세부 사항도 물리적 오류 없이 잘 표현되었습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "총구가 가방 천을 물리적으로 불가능하게 관통하고 있으며, 총을 쥔 손가락의 해부학적 구조가 훼손된 치명적 오류(Hard Violation)가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S84sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "리볼버 권총의 실린더, 프레임, 방아쇠울 등 금속 부품들의 형태가 비정상적으로 왜곡되고 융합되어 있습니다.",
     "fix_en": "Redraw the revolver to have clearly defined, mechanically accurate structural parts without any melted or distorted metal.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "깁스 밖으로 나온 손가락들이 제대로 형태를 갖추지 못하고 한 덩어리로 뭉개져 있습니다.",
     "fix_en": "Redraw the fingers emerging from the cast to be distinct, fully formed, and anatomically correct human fingers.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the revolver to have clearly defined, mechanically accurate structural parts without any melted or distorted metal.\n- Redraw the fingers emerging from the cast to be distinct, fully formed, and anatomically correct human fingers.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S84sh1"
  }
 },
 "S85sh4::bgfirst_bg": {
  "input_fingerprint": "edf7871f59ac97d4",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어구 창고 쪽을 향해 예리한 시선을 고정한 고산들의 굳은 얼굴 클로즈업.\n\nLOCATION (lock): On the village side of the island pier, facing the nearby fishing-gear warehouse.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At 고산들's face height and just off his eyeline, hold a firm static close-up with his face weighted slightly to one side and the warehouse direction left beyond the opposite frame edge. His rigid jaw and narrowed eyes carry the entire composition, while the shallow background prevents the unseen target of his suspicion from being visually resolved.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 선착장 옆 어구 창고 (Present beside the pier) — The warehouse lies just outside the frame along 고산들's sightline; used as Off-screen eyeline anchor that motivates 고산들's fixed stare.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어구 창고 쪽을 향해 예리한 시선을 고정한 고산들의 굳은 얼굴 클로즈업.\n\nLOCATION (lock): On the village side of the island pier, facing the nearby fishing-gear warehouse.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At 고산들's face height and just off his eyeline, hold a firm static close-up with his face weighted slightly to one side and the warehouse direction left beyond the opposite frame edge. His rigid jaw and narrowed eyes carry the entire composition, while the shallow background prevents the unseen target of his suspicion from being visually resolved.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 선착장 옆 어구 창고 (Present beside the pier) — The warehouse lies just outside the frame along 고산들's sightline; used as Off-screen eyeline anchor that motivates 고산들's fixed stare.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S85sh4__bgfirst_bg.png",
  "asset_id": "9fad90ac-4cff-4528-bd1f-fe4c175fc7b9",
  "input_asset_ids": [
   "c0a8eba1-98f6-4c9b-ab31-388c6f645794",
   "690ece37-1769-459c-a2ec-76a15158892c"
  ]
 },
 "S85sh4": {
  "input_fingerprint": "87c3fc29051d40b0",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어구 창고 쪽을 향해 예리한 시선을 고정한 고산들의 굳은 얼굴 클로즈업.\n\nLOCATION (lock): On the village side of the island pier, facing the nearby fishing-gear warehouse. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The fugitives remain hidden in the gear shed and coated in goat blood; Inwoo retains the .38 revolver.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어구 창고 쪽을 향해 예리한 시선을 고정한 고산들의 굳은 얼굴 클로즈업.\n\nLOCATION (lock): On the village side of the island pier, facing the nearby fishing-gear warehouse. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The fugitives remain hidden in the gear shed and coated in goat blood; Inwoo retains the .38 revolver.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어구 창고 쪽을 향해 예리한 시선을 고정한 고산들의 굳은 얼굴 클로즈업.\n\nLOCATION (lock): On the village side of the island pier, facing the nearby fishing-gear warehouse. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The fugitives remain hidden in the gear shed and coated in goat blood; Inwoo retains the .38 revolver.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S85sh4__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S85sh4.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:876576>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L18B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:876576>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "인물의 시선이 화면 우측의 창고 방향을 적절히 향하고 있으며, 굳은 얼굴의 클로즈업과 캐릭터 디자인을 충실히 반영했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "인물이 창고가 아닌 반대편 바다 쪽을 바라보고 있어 시선 방향에 대한 지문을 위반했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "인물의 시선이 화면 우측의 창고 방향을 적절히 향하고 있으며, 굳은 얼굴의 클로즈업과 캐릭터 디자인을 충실히 반영했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "인물이 창고가 아닌 반대편 바다 쪽을 바라보고 있어 시선 방향에 대한 지문을 위반했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "어구 창고가 위치한 화면 오른쪽으로 시선을 향하고 있어 '창고 쪽을 향해 고정한 시선'이라는 지문을 잘 충족합니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "어구 창고를 등지고 반대편 바다 쪽을 바라보고 있어 시선 방향이 지문의 요구사항과 상충됩니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "어구 창고가 위치한 화면 오른쪽으로 시선을 향하고 있어 '창고 쪽을 향해 고정한 시선'이라는 지문을 잘 충족합니다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "어구 창고를 등지고 반대편 바다 쪽을 바라보고 있어 시선 방향이 지문의 요구사항과 상충됩니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 9,
     "B": 15
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 9,
   "B": 15
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "인물의 시선이 화면 우측의 창고 방향을 적절히 향하고 있으며, 굳은 얼굴의 클로즈업과 캐릭터 디자인을 충실히 반영했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "인물이 창고가 아닌 반대편 바다 쪽을 바라보고 있어 시선 방향에 대한 지문을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L18B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "고산들이 배경에 있는 어구 창고를 향하지 않고 다른 방향을 보고 있습니다.",
     "fix_en": "Turn his head and eyes to gaze directly at the fishing-gear warehouse in the background.",
     "needs_regeneration": false,
     "unfixable": true
    }
   ]
  },
  "fix_skipped": true,
  "fix_skip_reason": "all_issues_unfixable",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S85sh4__bgfirst_bg.png",
   "bg_asset_id": "9fad90ac-4cff-4528-bd1f-fe4c175fc7b9",
   "bg_record_key": "S85sh4::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S85sh6": {
  "input_fingerprint": "b9721ee4aa2e1591",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 햇살이 부서지는 푸른 바다 위, 선수를 들고 하얀 물보라를 강하게 뿜어내며 바다를 가르고 있는 경찰 순찰정의 mid-action 외부 전경.\n\nLOCATION (lock): Across the bright sea approaching the island pier, around a fast-moving police patrol boat. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "기준 장소를 정확히 재현하고 선수를 들며 물보라를 일으키는 순찰정의 역동적인 모습을 성공적으로 묘사했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "경찰정의 묘사는 무난하나, 기준 이미지에 존재하는 건물의 창문과 우측 구조물들이 누락되어 장소 일치도가 떨어집니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "우측의 물보라가 건물 앞쪽으로 투명하게 겹쳐 보이는 물리적으로 불가능한 렌더링 오류가 발생하여 실격에 해당합니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L18B01.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "우측 표지판과 경찰정 측면에 텍스트와 의미 없는 문자가 생성되어 'no text' 지시를 위반했습니다.",
     "fix_en": "Remove all text and logos from the side of the boat and the sign on the wooden structure to the right.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all text and logos from the side of the boat and the sign on the wooden structure to the right.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S86sh7::bgfirst_bg": {
  "input_fingerprint": "e1e00defff996e4d",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 경찰 순찰정 갑판 위에 선 정인우가 갑판 아래에 있는 김지후(강수리영)와 강민숙(홍은미)의 뻗은 손을 꽉 쥐고 위로 끌어당기며 팔 근육이 팽팽하게 솟은 mid-action 찰나.\n\nLOCATION (lock): At the patrol boat's open deck edge beside the island pier during the escape.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From just outside the patrol boat at deck height, crane with a downward angle into a tight medium view of 정인우 bracing above while both women reach up from below. His two locked grips form a tense central triangle with their outstretched arms, while the deck edge cuts horizontally behind his forearms and preserves the difference in elevation.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 순찰정 갑판 가장자리 (Being used for emergency boarding) — Its outer side faces the camera, with the deck surface visible above it; used as Horizontal elevation divider between 정인우 and the two boarding women; 순찰정 갑판 (Occupied during the hurried escape); used as Immediate destination above the women and footing beneath 정인우.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 경찰 순찰정 갑판 위에 선 정인우가 갑판 아래에 있는 김지후(강수리영)와 강민숙(홍은미)의 뻗은 손을 꽉 쥐고 위로 끌어당기며 팔 근육이 팽팽하게 솟은 mid-action 찰나.\n\nLOCATION (lock): At the patrol boat's open deck edge beside the island pier during the escape.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From just outside the patrol boat at deck height, crane with a downward angle into a tight medium view of 정인우 bracing above while both women reach up from below. His two locked grips form a tense central triangle with their outstretched arms, while the deck edge cuts horizontally behind his forearms and preserves the difference in elevation.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 순찰정 갑판 가장자리 (Being used for emergency boarding) — Its outer side faces the camera, with the deck surface visible above it; used as Horizontal elevation divider between 정인우 and the two boarding women; 순찰정 갑판 (Occupied during the hurried escape); used as Immediate destination above the women and footing beneath 정인우.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S86sh7__bgfirst_bg.png",
  "asset_id": "43ac00a1-5c47-4afd-b1ce-821e46889599",
  "input_asset_ids": [
   "e67f324c-a6c4-4414-9491-9a8294f859c6",
   "e0e2d253-d296-466a-84b2-921275add8c9"
  ]
 },
 "S86sh7": {
  "input_fingerprint": "a9b90b5df6a20e89",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 경찰 순찰정 갑판 위에 선 정인우가 갑판 아래에 있는 김지후(강수리영)와 강민숙(홍은미)의 뻗은 손을 꽉 쥐고 위로 끌어당기며 팔 근육이 팽팽하게 솟은 mid-action 찰나.\n\nLOCATION (lock): At the patrol boat's open deck edge beside the island pier during the escape. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu, Minsuk, and Inwoo remain coated in goat blood as they board; Inwoo retains the .38 revolver and still has his injured arm cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상); 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비); 강민숙(홍은미) (한국인 40대 후반 여성, 갸름한 얼굴, 단정한 검은 머리, 귀티 나는 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 경찰 순찰정 갑판 위에 선 정인우가 갑판 아래에 있는 김지후(강수리영)와 강민숙(홍은미)의 뻗은 손을 꽉 쥐고 위로 끌어당기며 팔 근육이 팽팽하게 솟은 mid-action 찰나.\n\nLOCATION (lock): At the patrol boat's open deck edge beside the island pier during the escape. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu, Minsuk, and Inwoo remain coated in goat blood as they board; Inwoo retains the .38 revolver and still has his injured arm cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상); 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비); 강민숙(홍은미) (한국인 40대 후반 여성, 갸름한 얼굴, 단정한 검은 머리, 귀티 나는 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 경찰 순찰정 갑판 위에 선 정인우가 갑판 아래에 있는 김지후(강수리영)와 강민숙(홍은미)의 뻗은 손을 꽉 쥐고 위로 끌어당기며 팔 근육이 팽팽하게 솟은 mid-action 찰나.\n\nLOCATION (lock): At the patrol boat's open deck edge beside the island pier during the escape. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu, Minsuk, and Inwoo remain coated in goat blood as they board; Inwoo retains the .38 revolver and still has his injured arm cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상); 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비); 강민숙(홍은미) (한국인 40대 후반 여성, 갸름한 얼굴, 단정한 검은 머리, 귀티 나는 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S86sh7__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S86sh7.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:903809>"
    },
    {
     "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:844015>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L41B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:903809>"
    },
    {
     "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:844015>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "인우가 민숙의 손을 잡고 있지 않으며, 허리춤의 리볼버 형태와 위치가 물리적으로 어색하게 표현되어 지시문을 온전히 충족하지 못했습니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "인우가 지후와 민숙의 손을 모두 꽉 쥐고 당기는 근육질의 mid-action 순간을 피, 깁스, 리볼버 등 명시된 소품과 함께 가장 정확하게 구현했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "인우가 민숙의 손을 잡고 있지 않으며, 허리춤의 리볼버 형태와 위치가 물리적으로 어색하게 표현되어 지시문을 온전히 충족하지 못했습니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "인우가 지후와 민숙의 손을 모두 꽉 쥐고 당기는 근육질의 mid-action 순간을 피, 깁스, 리볼버 등 명시된 소품과 함께 가장 정확하게 구현했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "정인우가 두 사람의 손을 모두 잡고 끌어당기는 핵심 액션과 팽팽한 팔 근육을 지문대로 정확히 묘사했으며, 피 분장과 깁스, 리볼버 등 디테일도 잘 유지했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "정인우가 강민숙의 손을 잡지 않고 있어 '두 사람의 손을 꽉 쥐고'라는 핵심 액션 지시를 위반했으며, 김지후의 헤어스타일도 레퍼런스와 다릅니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "정인우가 두 사람의 손을 모두 잡고 끌어당기는 핵심 액션과 팽팽한 팔 근육을 지문대로 정확히 묘사했으며, 피 분장과 깁스, 리볼버 등 디테일도 잘 유지했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "정인우가 강민숙의 손을 잡지 않고 있어 '두 사람의 손을 꽉 쥐고'라는 핵심 액션 지시를 위반했으며, 김지후의 헤어스타일도 레퍼런스와 다릅니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 7,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 7,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "인우가 민숙의 손을 잡고 있지 않으며, 허리춤의 리볼버 형태와 위치가 물리적으로 어색하게 표현되어 지시문을 온전히 충족하지 못했습니다."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "인우가 지후와 민숙의 손을 모두 꽉 쥐고 당기는 근육질의 mid-action 순간을 피, 깁스, 리볼버 등 명시된 소품과 함께 가장 정확하게 구현했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L41B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903809>"
   },
   {
    "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:844015>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "정인우의 오른손과 김지후의 손이 맞잡은 부분이 해부학적으로 뭉개져 하나의 핏덩어리처럼 융합되어 있습니다.",
     "fix_en": "Redraw the grip between Inwoo's right arm and Jihu's hand to show two distinct, anatomically correct human hands firmly clasping each other.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "깁스를 한 정인우의 왼손과 강민숙이 맞잡은 부분의 손가락 형태가 비정상적으로 뭉개져 있습니다.",
     "fix_en": "Redraw the clasped hands between Inwoo's cast-covered left arm and Minsuk's arm so that both sets of fingers are anatomically correct and distinctly gripping.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "김지후의 상의가 캐릭터 레퍼런스의 짙은 녹색 스웨터가 아닌 밝은 색 셔츠로 렌더링되었습니다.",
     "fix_en": "Change Jihu's light-colored top to the dark green cable-knit sweater from her character reference, preserving the bloodstains.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "정인우가 캐릭터 레퍼런스에 있는 카키색 낚시 조끼와 올리브색 재킷을 입고 있지 않습니다.",
     "fix_en": "Clothe Inwoo in the olive green jacket and khaki fishing vest from his character reference, ensuring the bloodstains, arm cast, and tucked revolver remain visible.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the grip between Inwoo's right arm and Jihu's hand to show two distinct, anatomically correct human hands firmly clasping each other.\n- Redraw the clasped hands between Inwoo's cast-covered left arm and Minsuk's arm so that both sets of fingers are anatomically correct and distinctly gripping.\n- Change Jihu's light-colored top to the dark green cable-knit sweater from her character reference, preserving the bloodstains.\n- Clothe Inwoo in the olive green jacket and khaki fishing vest from his character reference, ensuring the bloodstains, arm cast, and tucked revolver remain visible.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S86sh7__bgfirst_bg.png",
   "bg_asset_id": "43ac00a1-5c47-4afd-b1ce-821e46889599",
   "bg_record_key": "S86sh7::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S86sh12": {
  "input_fingerprint": "8e411cb9692cd8fe",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 바닥에 널브러진 조 형사의 시체를 배경으로 한 채, 고산들이 순찰정을 향해 상체를 숙이고 한 발을 뻗어 짐승처럼 내달리는 mid-stride 전신.\n\nLOCATION (lock): Across the island pier between the fallen detective and the departing patrol boat. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the crowded pier, patrol boat, dead detective, daylight, and villagers rushing downhill. Exclude the boarding hand-pull action and show the large man sprinting toward the boat.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): His corpse is sprawled where it struck the ground, with the neck sharply twisted by the fatal blow and the arms and legs fallen slack against the dock.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Detective Jo's corpse is sprawled where it struck the ground, with the neck sharply twisted by the fatal blow and the arms and legs fallen slack against the dock. The escaping trio remain goat-blood stained, and Inwoo retains the revolver.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상); 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비); 강민숙(홍은미) (한국인 40대 후반 여성, 갸름한 얼굴, 단정한 검은 머리, 귀티 나는 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 8,
   "A": 6,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "상체를 숙이고 짐승처럼 내달리는 역동적인 액션(우선순위 2), 언덕을 내려오는 마을 사람들의 배치(우선순위 3), 시체의 꺾인 목 자세(우선순위 5)를 훌륭히 구현했으나 인우의 팔 붕대(우선순위 4)가 누락됨."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "인우의 팔뚝 붕대와 인물들의 외모는 훌륭하나, 짐승처럼 달리는 역동적인 자세와 시체의 목 꺾임이 부족하고 언덕을 내려오는 마을 사람들이 제대로 묘사되지 않음."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "지정된 인물 목록(지후, 인우, 민숙)에 없는 새로운 인물을 메인으로 등장시켜 프롬프트의 엄격한 인물 제한 규칙을 위반한 심각한 오류(Hard Violation)."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S86sh7_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1225321>"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1092285>"
   },
   {
    "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1269524>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "정인우의 오른쪽 허리춤에 지시된 리볼버 권총이 없고 검은색 무전기 같은 형태의 물건이 있습니다.",
     "fix_en": "Replace the black rectangular object on the running man's right waistband with the brown-gripped revolver from the reference image.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "배경에 쓰러진 조 형사 시체의 목이 치명타로 인해 심하게 꺾여 있어야 한다는 지시와 달리, 머리가 똑바로 위를 향해 놓여 있습니다.",
     "fix_en": "Edit the head and neck of the dead man lying in the background so that his neck is sharply twisted to the side.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "텍스트를 배제하라는 지시(No text anywhere)와 달리, 배경의 흰색 건물 간판과 순찰정 조타실 외벽에 텍스트가 포함되어 있습니다.",
     "fix_en": "Remove the blue text from the sign on the white building in the background and the text from the exterior wall of the patrol boat.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the black rectangular object on the running man's right waistband with the brown-gripped revolver from the reference image.\n- Edit the head and neck of the dead man lying in the background so that his neck is sharply twisted to the side.\n- Remove the blue text from the sign on the white building in the background and the text from the exterior wall of the patrol boat.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S86sh7"
  }
 },
 "S87sh15::bgfirst_bg": {
  "input_fingerprint": "0277bde98bc654c1",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 공포에 질려 울먹이는 김지후(강수리영)의 앞을 굳건한 표정의 강민숙(홍은미)이 떡하니 막아선 뒷모습.\n\nLOCATION (lock): On the exposed deck of the moving police patrol boat, between the attacker and the stern.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Directly behind 강민숙(홍은미) at shoulder height, hold her squared back across the central foreground as a human barrier, with 김지후(강수리영) visible beyond one shoulder in the stern area. The static medium frame lets 민숙's planted posture replace the preceding panic with a single immovable silhouette, while the threatening passage remains ahead and outside the frame.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 선수와 선미를 연결하는 통로 (Separates the approaching threat from the stern) — The passage recedes away from the camera beyond 강민숙; used as Depth corridor behind 민숙 that defines the direction of the threat; 순찰정 난간 (Beside the confrontation area) — Its inner side runs along the edge of the deck beside the characters; used as Lateral boundary preparing the following move toward the sacrifice; 선미 갑판 (Occupied by the terrified 수리영); used as Lower spatial plane supporting 수리영's defensive posture.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 공포에 질려 울먹이는 김지후(강수리영)의 앞을 굳건한 표정의 강민숙(홍은미)이 떡하니 막아선 뒷모습.\n\nLOCATION (lock): On the exposed deck of the moving police patrol boat, between the attacker and the stern.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Directly behind 강민숙(홍은미) at shoulder height, hold her squared back across the central foreground as a human barrier, with 김지후(강수리영) visible beyond one shoulder in the stern area. The static medium frame lets 민숙's planted posture replace the preceding panic with a single immovable silhouette, while the threatening passage remains ahead and outside the frame.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 선수와 선미를 연결하는 통로 (Separates the approaching threat from the stern) — The passage recedes away from the camera beyond 강민숙; used as Depth corridor behind 민숙 that defines the direction of the threat; 순찰정 난간 (Beside the confrontation area) — Its inner side runs along the edge of the deck beside the characters; used as Lateral boundary preparing the following move toward the sacrifice; 선미 갑판 (Occupied by the terrified 수리영); used as Lower spatial plane supporting 수리영's defensive posture.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S87sh15__bgfirst_bg.png",
  "asset_id": "7e3b6673-b152-4a90-a9cd-2edca392f548",
  "input_asset_ids": [
   "61bf7aa7-6e66-45e3-bb14-35143a2d7b40",
   "06b17b12-bc26-4394-94b6-7ce08d1a5bdc"
  ]
 },
 "S87sh15": {
  "input_fingerprint": "d79dfe189caf50a7",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 공포에 질려 울먹이는 김지후(강수리영)의 앞을 굳건한 표정의 강민숙(홍은미)이 떡하니 막아선 뒷모습.\n\nLOCATION (lock): On the exposed deck of the moving police patrol boat, between the attacker and the stern. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul bears the gunshot wound in his shoulder; Detective Shin lies gravely wounded from the neck bite, and Inwoo has been thrown down after losing the revolver.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 공포에 질려 울먹이는 김지후(강수리영)의 앞을 굳건한 표정의 강민숙(홍은미)이 떡하니 막아선 뒷모습.\n\nLOCATION (lock): On the exposed deck of the moving police patrol boat, between the attacker and the stern. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul bears the gunshot wound in his shoulder; Detective Shin lies gravely wounded from the neck bite, and Inwoo has been thrown down after losing the revolver.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 공포에 질려 울먹이는 김지후(강수리영)의 앞을 굳건한 표정의 강민숙(홍은미)이 떡하니 막아선 뒷모습.\n\nLOCATION (lock): On the exposed deck of the moving police patrol boat, between the attacker and the stern. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul bears the gunshot wound in his shoulder; Detective Shin lies gravely wounded from the neck bite, and Inwoo has been thrown down after losing the revolver.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S87sh15__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S87sh15.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_village_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 샷 구도와 부상자들의 디테일(어깨 상처, 떨어진 권총)을 사실적으로 구현했으나, 레퍼런스 장소 구조물이 생략되었습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "레퍼런스 장소를 배경에 반영했으나, 좌측 인물이 배 밖 허공에 떠 있고 우측 인물들의 신체가 엉켜있는 물리적 오류(Hard Violation)가 있습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 샷 구도와 부상자들의 디테일(어깨 상처, 떨어진 권총)을 사실적으로 구현했으나, 레퍼런스 장소 구조물이 생략되었습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "레퍼런스 장소를 배경에 반영했으나, 좌측 인물이 배 밖 허공에 떠 있고 우측 인물들의 신체가 엉켜있는 물리적 오류(Hard Violation)가 있습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "샷 텍스트의 구도와 감정 표현을 정확히 구현했고, 레퍼런스의 해안 구조물과 지시된 인물들의 상태까지 충실히 반영했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "전경의 인물 구도는 적절하나, 레퍼런스 이미지의 해안 구조물이 배경에 전혀 반영되지 않았고 지정된 주변 인물 일부가 누락되었습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "샷 텍스트의 구도와 감정 표현을 정확히 구현했고, 레퍼런스의 해안 구조물과 지시된 인물들의 상태까지 충실히 반영했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "전경의 인물 구도는 적절하나, 레퍼런스 이미지의 해안 구조물이 배경에 전혀 반영되지 않았고 지정된 주변 인물 일부가 누락되었습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 11
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 11,
   "B": 11
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지정된 샷 구도와 부상자들의 디테일(어깨 상처, 떨어진 권총)을 사실적으로 구현했으나, 레퍼런스 장소 구조물이 생략되었습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "레퍼런스 장소를 배경에 반영했으나, 좌측 인물이 배 밖 허공에 떠 있고 우측 인물들의 신체가 엉켜있는 물리적 오류(Hard Violation)가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_village_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "왼쪽 서 있는 남자의 오른손에 레이아웃 스케치에 그려진 권총이 없습니다.",
     "fix_en": "Place a handgun in the right hand of the standing man on the left, hanging down by his side.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "오른쪽에 쓰러진 남자의 자세가 스케치(왼쪽 팔꿈치로 상체를 지탱하고 뒤를 돌아보는 자세)와 다르게 완전히 누워있습니다.",
     "fix_en": "Re-render the fallen man on the right so he is propped up on his left elbow and looking back over his shoulder to the left.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Place a handgun in the right hand of the standing man on the left, hanging down by his side.\n- Re-render the fallen man on the right so he is propped up on his left elbow and looking back over his shoulder to the left.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S87sh15__bgfirst_bg.png",
   "bg_asset_id": "7e3b6673-b152-4a90-a9cd-2edca392f548",
   "bg_record_key": "S87sh15::bgfirst_bg",
   "chain_winner": true,
   "authority": "seed_bg"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S87sh19": {
  "input_fingerprint": "aaca73d5263b934b",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 강민숙(홍은미)이 고산들을 꽉 끌어안은 채 순찰정 난간 너머로 함께 몸을 던져 허공에 떠 있는 순간.\n\nLOCATION (lock): At the patrol boat's side rail above the open sea. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the moving patrol boat, wet deck, rails, open sea, and continuing combat positions. Exclude the mother's protective stand and show her carrying the man over the rail.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul retains his shoulder gunshot wound as Minsuk locks him in an embrace and carries both of them over the rail.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 8,
   "A": 3,
   "B": 2
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "난간 너머로 몸을 던져 허공에 떠 있는 샷의 구도와 어깨 총상 등 디테일을 가장 사실적으로 완벽하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "인물들이 배경과 비율이 맞지 않게 합성(콜라주)된 것처럼 보여 물리적 자연스러움이 떨어집니다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "난간 너머 바다가 아닌 갑판 위에서 공중 부양하고 있어 샷의 핵심적인 공간 연출을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S87sh15_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1215674>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "여자의 팔이 3개로 묘사되었습니다. 남자의 목을 감싼 팔과 허리를 감싼 팔 외에, 어깨 상처 바로 아래쪽 등 중간을 쥐고 있는 여분의 손과 소매가 있습니다.",
     "fix_en": "Remove the extra hand and grey sleeve clutching the middle of the man's back, keeping only the woman's top and bottom arms."
    },
    {
     "issue_ko": "오른쪽 난간에 얹혀 있는 회색 바지와 흰색 운동화를 신은 다리가 인체 구조상 불가능한 위치(남자의 다리 부근)에서 뻗어 나와 신체와 연결되어 있지 않습니다.",
     "fix_en": "Remove the anatomically disconnected grey leg and white sneaker resting on the boat railing."
    },
    {
     "issue_ko": "고산들(남자)이 레퍼런스 이미지에서 착용한 검은색 비니가 누락되었으며, 얼굴에 있는 상처와 핏자국이 반영되지 않았습니다.",
     "fix_en": "Add the black beanie to the man's head and match the facial cuts and blood exactly as shown in his character reference."
    },
    {
     "issue_ko": "순찰정의 두꺼운 외벽, 계단식 갑판, 난간 구조가 레퍼런스(평평한 갑판 가장자리의 얇은 파이프 난간)와 전혀 일치하지 않습니다.",
     "fix_en": "Rebuild the boat's deck and railing to match the flat deck and simple pipe railing established in the reference image.",
     "unfixable": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the extra hand and grey sleeve clutching the middle of the man's back, keeping only the woman's top and bottom arms.\n- Remove the anatomically disconnected grey leg and white sneaker resting on the boat railing.\n- Add the black beanie to the man's head and match the facial cuts and blood exactly as shown in his character reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S87sh15"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S87sh23": {
  "input_fingerprint": "3a12524c5c783e9d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 정인우에게 붙들린 채 멀어지는 바다의 포말을 바라보며 절망적으로 오열하는 김지후(강수리영)의 클로즈업.\n\nLOCATION (lock): At the stern rail of the departing patrol boat, looking back over the foaming sea. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the moving boat, rear rail, foamy wake, and receding splash site. Exclude the midair fall and show the daughter restrained while looking back.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk and Sandeul remain submerged and out of sight; wounded Detective Shin and the fallen Inwoo remain aboard the patrol boat.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 4,
   "B": 2
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "요구된 클로즈업 샷에 가장 부합하며, 인물에게 '붙들린 채 오열하는' 구속 상태와 감정을 오류 없이 정확하게 연출했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "클로즈업 지시를 어기고 프레임이 너무 넓게 잡혔으며, 단순히 어깨에 손을 올린 정도로 '붙들린 채'라는 강한 구속의 행동 묘사가 부족합니다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "여성을 감싸는 팔과 몸통의 연결이 누락되거나 왜곡되는 등 심각하고 불가능한 해부학적 오류(Hard Violation)가 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S87sh19_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1226689>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "여자의 시선이 바다의 포말(왼쪽)을 향하지 않고 오른쪽 아래를 향하고 있습니다.",
     "fix_en": "Redirect the woman's eyes to the left so she is looking toward the white sea foam in the background.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "이전 컷에서 남자의 오른쪽 팔뚝에 있던 붉은 핏자국(상처)이 사라졌습니다.",
     "fix_en": "Add the bloody red wound to the man's right upper arm on the dark grey sleeve wrapping around the woman's chest.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "남자의 재킷에 이전 컷에는 없던 견장, 가슴 주머니, 소매 주머니가 생겼습니다.",
     "fix_en": "Remove the epaulettes, breast pockets, and sleeve patch from the man's dark grey jacket to match the plain style of the reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redirect the woman's eyes to the left so she is looking toward the white sea foam in the background.\n- Add the bloody red wound to the man's right upper arm on the dark grey sleeve wrapping around the woman's chest.\n- Remove the epaulettes, breast pockets, and sleeve patch from the man's dark grey jacket to match the plain style of the reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S87sh19"
  }
 },
 "S88sh3::bgfirst_bg": {
  "input_fingerprint": "3bab8a709847eefa",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 금월도 선착장에서 붉은색 무녀복을 입은 강영란(흰 머리 여자), 고산들, 금월도 마을 주민들이 배를 향해 고정된 자세로 서 있는 전경.\n\nLOCATION (lock): On the island's opposite landing shore, facing the patrol boat as it returns toward land.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the patrol boat's forward section at eye level, hold a static long-distance wide composition past the bow toward the dock, where 강영란(흰 머리 여자), 고산들, and 금월도 마을 주민 form an unnaturally fixed horizontal line facing the vessel. The bow occupies only the lower edge as a scale cue, while 강영란's red shaman garment marks the central human anchor among the waiting figures.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 순찰정 선수 (Approaching 금월도) — Its forward-facing upper edge points toward the dock at the bottom of frame; used as Foreground scale cue establishing the camera aboard the approaching vessel; 금월도 선착장 (Occupied by 강영란, 고산들, and the villagers) — Its boat-facing edge is visible beyond the patrol boat's bow; used as Long-distance platform supporting the fixed group tableau; 금월도 육지 (Visible beyond the dock); used as Background confirmation that the vessel has returned to the island.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 금월도 선착장에서 붉은색 무녀복을 입은 강영란(흰 머리 여자), 고산들, 금월도 마을 주민들이 배를 향해 고정된 자세로 서 있는 전경.\n\nLOCATION (lock): On the island's opposite landing shore, facing the patrol boat as it returns toward land.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the patrol boat's forward section at eye level, hold a static long-distance wide composition past the bow toward the dock, where 강영란(흰 머리 여자), 고산들, and 금월도 마을 주민 form an unnaturally fixed horizontal line facing the vessel. The bow occupies only the lower edge as a scale cue, while 강영란's red shaman garment marks the central human anchor among the waiting figures.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 순찰정 선수 (Approaching 금월도) — Its forward-facing upper edge points toward the dock at the bottom of frame; used as Foreground scale cue establishing the camera aboard the approaching vessel; 금월도 선착장 (Occupied by 강영란, 고산들, and the villagers) — Its boat-facing edge is visible beyond the patrol boat's bow; used as Long-distance platform supporting the fixed group tableau; 금월도 육지 (Visible beyond the dock); used as Background confirmation that the vessel has returned to the island.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S88sh3__bgfirst_bg.png",
  "asset_id": "40779df0-94ba-43f5-a58c-88a67ee67e0c",
  "input_asset_ids": [
   "5014b414-b0dc-4965-a548-65b31780b4b0",
   "5920c8a3-9754-40bd-96db-d86a0391e5df"
  ]
 },
 "S88sh3": {
  "input_fingerprint": "312097f3f7aa3099",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 금월도 선착장에서 붉은색 무녀복을 입은 강영란(흰 머리 여자), 고산들, 금월도 마을 주민들이 배를 향해 고정된 자세로 서 있는 전경.\n\nLOCATION (lock): On the island's opposite landing shore, facing the patrol boat as it returns toward land. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul has returned to the island after the fall into the sea and retains his injuries; Jihu’s pendant remains with Youngran’s ritual preparations.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매); 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 금월도 선착장에서 붉은색 무녀복을 입은 강영란(흰 머리 여자), 고산들, 금월도 마을 주민들이 배를 향해 고정된 자세로 서 있는 전경.\n\nLOCATION (lock): On the island's opposite landing shore, facing the patrol boat as it returns toward land. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul has returned to the island after the fall into the sea and retains his injuries; Jihu’s pendant remains with Youngran’s ritual preparations.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매); 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 금월도 선착장에서 붉은색 무녀복을 입은 강영란(흰 머리 여자), 고산들, 금월도 마을 주민들이 배를 향해 고정된 자세로 서 있는 전경.\n\nLOCATION (lock): On the island's opposite landing shore, facing the patrol boat as it returns toward land. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul has returned to the island after the fall into the sea and retains his injuries; Jihu’s pendant remains with Youngran’s ritual preparations.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매); 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S88sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S88sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:876576>"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:902182>"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_dock_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:876576>"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:902182>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "카메라가 배 위에 위치하여 인물들이 배를 향해 서 있는 구도를 정확히 연출했으며, 레퍼런스의 장소 구조와 복장을 훌륭하게 반영했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "배가 인물들 뒤에 위치하여 '배를 향해 서 있는'이라는 핵심 지시문을 위반했으며, 주요 인물의 복장 디테일이 레퍼런스와 다릅니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "카메라가 배 위에 위치하여 인물들이 배를 향해 서 있는 구도를 정확히 연출했으며, 레퍼런스의 장소 구조와 복장을 훌륭하게 반영했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "배가 인물들 뒤에 위치하여 '배를 향해 서 있는'이라는 핵심 지시문을 위반했으며, 주요 인물의 복장 디테일이 레퍼런스와 다릅니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "접근하는 배의 시점에서 선착장에 서서 배를 응시하는 인물들의 구도를 지시문과 로케이션 레퍼런스에 맞게 완벽히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "화면 뒤쪽에 배가 배치되어 인물들이 배를 등지고 서 있으므로 '배를 향해 서 있는' 핵심 연출 지시를 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "접근하는 배의 시점에서 선착장에 서서 배를 응시하는 인물들의 구도를 지시문과 로케이션 레퍼런스에 맞게 완벽히 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "화면 뒤쪽에 배가 배치되어 인물들이 배를 등지고 서 있으므로 '배를 향해 서 있는' 핵심 연출 지시를 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 6
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 16,
   "B": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "카메라가 배 위에 위치하여 인물들이 배를 향해 서 있는 구도를 정확히 연출했으며, 레퍼런스의 장소 구조와 복장을 훌륭하게 반영했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "배가 인물들 뒤에 위치하여 '배를 향해 서 있는'이라는 핵심 지시문을 위반했으며, 주요 인물의 복장 디테일이 레퍼런스와 다릅니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_dock_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:902182>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "고산들(왼쪽에서 4번째)은 바다에 빠졌던 상태를 나타내는 부상이나 옷이 젖고 더러워진 흔적 없이 완전히 깨끗한 모습입니다.",
     "fix_en": "Modify the man in black (4th from the left) to have visible dirt, wet clothing, and injuries, reflecting his recent fall into the sea.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "레이아웃 스케치에서 소년은 오른쪽에서 두 번째(11번째 위치)에 있어야 하나 세 번째에 배치되었으며, 소년 레퍼런스의 복장(모자, 장화 등)이 옆에 있는 성인 남성에게 잘못 적용되었습니다.",
     "fix_en": "Make the 3rd person from the right a tall adult man, and make the 2nd person from the right a young boy wearing the exact cap, blue vest, brown pants, and green boots from the character reference.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "오른쪽에서 4번째(왼쪽에서 9번째) 남성은 레이아웃 스케치에 지정된 대로 팔짱을 끼고 있어야 하지만 팔을 내리고 있습니다.",
     "fix_en": "Change the pose of the man in the light blue shirt (4th from the right) so that his arms are crossed over his chest.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "강영란(붉은 옷을 입은 여자) 주변에 있어야 할 무속 의식 준비물과 지후의 펜던트가 명확하게 보이지 않습니다.",
     "fix_en": "Add recognizable ritual preparation items and a distinct pendant on the ground immediately next to the woman in the red dress.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Modify the man in black (4th from the left) to have visible dirt, wet clothing, and injuries, reflecting his recent fall into the sea.\n- Make the 3rd person from the right a tall adult man, and make the 2nd person from the right a young boy wearing the exact cap, blue vest, brown pants, and green boots from the character reference.\n- Change the pose of the man in the light blue shirt (4th from the right) so that his arms are crossed over his chest.\n- Add recognizable ritual preparation items and a distinct pendant on the ground immediately next to the woman in the red dress.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S88sh3__bgfirst_bg.png",
   "bg_asset_id": "40779df0-94ba-43f5-a58c-88a67ee67e0c",
   "bg_record_key": "S88sh3::bgfirst_bg",
   "chain_winner": true,
   "authority": "seed_bg"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S88sh6::bgfirst_bg": {
  "input_fingerprint": "0765ef92a254ea69",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 조타실 안의 구창모가 총구를 밖의 정인우를 향해 정확히 겨누고 위협하는 찰나.\n\nLOCATION (lock): Inside the patrol boat's wheelhouse, steering toward the island landing while aiming through the cabin opening.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Inside the wheelhouse at shoulder height, dolly in obliquely along 구창모's extended gun arm so his three-quarter figure occupies one side and 정인우 remains visible outside in the opposing depth plane. The barrel creates an unmistakable diagonal between them, while 정인우's arrested posture and partial profile make the threat legible without centering either man frontally.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 조타실 (Occupied and controlled by 구창모) — The interior faces outward toward 정인우 on the deck; used as Interior spatial frame around 구창모 and the origin point of the threat; 조타실 키 (Under 구창모's control) — The operating side is visible near 구창모 within the wheelhouse; used as Secondary evidence identifying 구창모 as the person controlling the patrol boat; 구창모의 총 (Raised and aimed at 정인우) — Its barrel points from the wheelhouse toward 정인우 outside; used as Strong diagonal focal prop linking aggressor and hostage.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 조타실 안의 구창모가 총구를 밖의 정인우를 향해 정확히 겨누고 위협하는 찰나.\n\nLOCATION (lock): Inside the patrol boat's wheelhouse, steering toward the island landing while aiming through the cabin opening.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Inside the wheelhouse at shoulder height, dolly in obliquely along 구창모's extended gun arm so his three-quarter figure occupies one side and 정인우 remains visible outside in the opposing depth plane. The barrel creates an unmistakable diagonal between them, while 정인우's arrested posture and partial profile make the threat legible without centering either man frontally.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 조타실 (Occupied and controlled by 구창모) — The interior faces outward toward 정인우 on the deck; used as Interior spatial frame around 구창모 and the origin point of the threat; 조타실 키 (Under 구창모's control) — The operating side is visible near 구창모 within the wheelhouse; used as Secondary evidence identifying 구창모 as the person controlling the patrol boat; 구창모의 총 (Raised and aimed at 정인우) — Its barrel points from the wheelhouse toward 정인우 outside; used as Strong diagonal focal prop linking aggressor and hostage.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S88sh6__bgfirst_bg.png",
  "asset_id": "5b0aef52-668e-40d4-b126-9747c56cd5a0",
  "input_asset_ids": [
   "6fe3a12c-4fd1-4e91-b84c-8953cfcb5b4e",
   "5920c8a3-9754-40bd-96db-d86a0391e5df"
  ]
 },
 "S88sh6": {
  "input_fingerprint": "de981f63b669ef81",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 조타실 안의 구창모가 총구를 밖의 정인우를 향해 정확히 겨누고 위협하는 찰나.\n\nLOCATION (lock): Inside the patrol boat's wheelhouse, steering toward the island landing while aiming through the cabin opening. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Changmo controls the patrol boat and keeps his gun trained on Inwoo; Sandeul is back ashore after the immersion.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상); 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 조타실 안의 구창모가 총구를 밖의 정인우를 향해 정확히 겨누고 위협하는 찰나.\n\nLOCATION (lock): Inside the patrol boat's wheelhouse, steering toward the island landing while aiming through the cabin opening. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Changmo controls the patrol boat and keeps his gun trained on Inwoo; Sandeul is back ashore after the immersion.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상); 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 조타실 안의 구창모가 총구를 밖의 정인우를 향해 정확히 겨누고 위협하는 찰나.\n\nLOCATION (lock): Inside the patrol boat's wheelhouse, steering toward the island landing while aiming through the cabin opening. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Changmo controls the patrol boat and keeps his gun trained on Inwoo; Sandeul is back ashore after the immersion.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 금월도 마을 주민 (한국인 남녀노소, 검은 머리, 평범한 섬 주민 얼굴, 햇볕에 그을린 인상); 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S88sh6__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S88sh6.png"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1385833>"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1226689>"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_dock_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1385833>"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1226689>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "카메라가 조타실 외부에 위치한 점은 아쉽지만, 두 인물의 얼굴과 복장, 선착장 표지판 등 참조 이미지의 디테일을 매우 충실하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "총구와 시선을 연결하는 검은색 지시선이 화면에 그대로 노출되는 치명적인 오류(Hard Violation)가 있어 가장 낮은 평가를 받습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "카메라가 조타실 외부에 위치한 점은 아쉽지만, 두 인물의 얼굴과 복장, 선착장 표지판 등 참조 이미지의 디테일을 매우 충실하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "총구와 시선을 연결하는 검은색 지시선이 화면에 그대로 노출되는 치명적인 오류(Hard Violation)가 있어 가장 낮은 평가를 받습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "조타실 내부에서 창밖의 인물을 향해 총을 겨누는 샷의 구도와 지정된 인물 및 배경 레퍼런스를 충실하게 반영했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "인물의 시선을 나타내는 검은색 지시선(마커)이 이미지에 그대로 노출되는 치명적인 규정 위반(Hard Violation)이 있습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "조타실 내부에서 창밖의 인물을 향해 총을 겨누는 샷의 구도와 지정된 인물 및 배경 레퍼런스를 충실하게 반영했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "인물의 시선을 나타내는 검은색 지시선(마커)이 이미지에 그대로 노출되는 치명적인 규정 위반(Hard Violation)이 있습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "카메라가 조타실 외부에 위치한 점은 아쉽지만, 두 인물의 얼굴과 복장, 선착장 표지판 등 참조 이미지의 디테일을 매우 충실하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "총구와 시선을 연결하는 검은색 지시선이 화면에 그대로 노출되는 치명적인 오류(Hard Violation)가 있어 가장 낮은 평가를 받습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_dock_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 금월도 마을 주민: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1385833>"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1226689>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "화면 내 모든 텍스트 배제를 지시했으나 밖의 선착장 표지판에 글자가 있습니다.",
     "fix_en": "Remove all text and characters from the sign on the pier.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "지문에서 배를 조종하는(steering) 상태를 요구했으나, 인물이 조타기를 잡고 있지 않습니다.",
     "fix_en": "Raise the aiming man's left arm and place his hand firmly on the wooden steering wheel.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "밖의 인물을 정확히 겨누고 위협해야 하나, 총을 든 남자의 시선이 엉뚱하게 아래쪽을 향하고 있습니다.",
     "fix_en": "Redirect the aiming man's eyes upward so he is looking directly at the man in the suit standing outside.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all text and characters from the sign on the pier.\n- Raise the aiming man's left arm and place his hand firmly on the wooden steering wheel.\n- Redirect the aiming man's eyes upward so he is looking directly at the man in the suit standing outside.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S88sh6__bgfirst_bg.png",
   "bg_asset_id": "5b0aef52-668e-40d4-b126-9747c56cd5a0",
   "bg_record_key": "S88sh6::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S89sh1::bgfirst_bg": {
  "input_fingerprint": "36fbeed799522a81",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 물살을 일으키며 다가오는 중인 순찰정을 향해 만족스러운 미소를 머금은 채 시선을 고정하고 있는 강영란(흰 머리 여자)의 얼굴 클로즈업.\n\nLOCATION (lock): On the rocky island landing directly before the approaching patrol boat.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level, hold a close frontal three-quarter view just beyond 강영란(흰 머리 여자)'s eyeline, keeping her face slightly off-center while the direction of the unseen patrol boat remains beyond the lens. The camera stays static as her held smile and unwavering off-screen gaze occupy most of the frame, with only a narrow trace of the rocky shore behind her.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 금월도 갯바위 (Only a narrow portion is visible behind 강영란(흰 머리 여자)); used as A narrow background fragment that preserves the coastal location without distracting from the expression.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 물살을 일으키며 다가오는 중인 순찰정을 향해 만족스러운 미소를 머금은 채 시선을 고정하고 있는 강영란(흰 머리 여자)의 얼굴 클로즈업.\n\nLOCATION (lock): On the rocky island landing directly before the approaching patrol boat.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level, hold a close frontal three-quarter view just beyond 강영란(흰 머리 여자)'s eyeline, keeping her face slightly off-center while the direction of the unseen patrol boat remains beyond the lens. The camera stays static as her held smile and unwavering off-screen gaze occupy most of the frame, with only a narrow trace of the rocky shore behind her.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 금월도 갯바위 (Only a narrow portion is visible behind 강영란(흰 머리 여자)); used as A narrow background fragment that preserves the coastal location without distracting from the expression.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S89sh1__bgfirst_bg.png",
  "asset_id": "8a34f925-81f4-4f6b-88f7-fc4db29a6054",
  "input_asset_ids": [
   "bb32291b-15f9-42d4-ba82-e5738ce3d573",
   "29a2c34f-c295-47e6-a051-3fe950ac4bd2"
  ]
 },
 "S89sh1": {
  "input_fingerprint": "3da0358d25c73c9b",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 물살을 일으키며 다가오는 중인 순찰정을 향해 만족스러운 미소를 머금은 채 시선을 고정하고 있는 강영란(흰 머리 여자)의 얼굴 클로즈업.\n\nLOCATION (lock): On the rocky island landing directly before the approaching patrol boat. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul remains ashore despite having fallen into the sea, and Changmo continues bringing the patrol boat back toward Geumwoldo.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 물살을 일으키며 다가오는 중인 순찰정을 향해 만족스러운 미소를 머금은 채 시선을 고정하고 있는 강영란(흰 머리 여자)의 얼굴 클로즈업.\n\nLOCATION (lock): On the rocky island landing directly before the approaching patrol boat. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul remains ashore despite having fallen into the sea, and Changmo continues bringing the patrol boat back toward Geumwoldo.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 물살을 일으키며 다가오는 중인 순찰정을 향해 만족스러운 미소를 머금은 채 시선을 고정하고 있는 강영란(흰 머리 여자)의 얼굴 클로즈업.\n\nLOCATION (lock): On the rocky island landing directly before the approaching patrol boat. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul remains ashore despite having fallen into the sea, and Changmo continues bringing the patrol boat back toward Geumwoldo.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S89sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S89sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L38B01.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "흰 머리 여자의 얼굴 클로즈업과 만족스러운 미소가 잘 표현되었으며, 다가오는 순찰정과 물살이 시각적으로 더 돋보입니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "인물의 표정과 클로즈업 샷은 훌륭하나, 배경의 순찰정이 너무 멀어 '물살을 일으키며 다가오는' 느낌이 상대적으로 약합니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "흰 머리 여자의 얼굴 클로즈업과 만족스러운 미소가 잘 표현되었으며, 다가오는 순찰정과 물살이 시각적으로 더 돋보입니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "인물의 표정과 클로즈업 샷은 훌륭하나, 배경의 순찰정이 너무 멀어 '물살을 일으키며 다가오는' 느낌이 상대적으로 약합니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "요구된 얼굴 클로즈업과 표정을 잘 담아냈으며, 인물의 시선 방향과 배경의 다가오는 순찰정 위치를 일치시켜 관계를 정확히 연출했습니다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "클로즈업 샷의 틀은 맞추었으나, 인물이 바라보는 방향과 순찰정의 위치가 반대로 배치되어 배를 향해 시선을 고정한다는 지시문을 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "요구된 얼굴 클로즈업과 표정을 잘 담아냈으며, 인물의 시선 방향과 배경의 다가오는 순찰정 위치를 일치시켜 관계를 정확히 연출했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "클로즈업 샷의 틀은 맞추었으나, 인물이 바라보는 방향과 순찰정의 위치가 반대로 배치되어 배를 향해 시선을 고정한다는 지시문을 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 13,
     "B": 12
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 13,
   "B": 12
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "흰 머리 여자의 얼굴 클로즈업과 만족스러운 미소가 잘 표현되었으며, 다가오는 순찰정과 물살이 시각적으로 더 돋보입니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "인물의 표정과 클로즈업 샷은 훌륭하나, 배경의 순찰정이 너무 멀어 '물살을 일으키며 다가오는' 느낌이 상대적으로 약합니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L38B01.png"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S89sh1__bgfirst_bg.png",
   "bg_asset_id": "8a34f925-81f4-4f6b-88f7-fc4db29a6054",
   "bg_record_key": "S89sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S90sh1": {
  "input_fingerprint": "fceeeeffefbd884d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk, reddish sunset.\n\nSHOT TEXT (authoritative, Korean): 붉으스름한 노을빛이 스며드는 텅 빈 반월 식당 내부의 적막한 전경.\n\nLOCATION (lock): Inside the empty island restaurant at sunset, filled with soft reddish light entering through its windows. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 9,
   "B": 6,
   "A": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 9,
    "verdict_ko": "텅 빈 반월 식당의 적막한 분위기와 붉은 노을빛을 완벽히 연출했으며, 레퍼런스의 선박 구조물 디테일(문, 환기구 상자, 창문)을 내부 공간에 탁월하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "붉은 노을과 텅 빈 실내라는 조건은 충족했으나, 레퍼런스에 있는 고유한 구조물(경사진 상단 창문, 환기구 상자 등)의 형태와 비율을 대부분 누락했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "노을빛이 드는 빈 공간은 표현되었으나, 지정된 위치 레퍼런스의 구조적 특징(조타실 외형)이 완전히 배제된 채 평범한 방으로 묘사되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L38B02.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_police_patrol_boat_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트에서 텍스트 생성을 금지했으나 화면 상단과 벽면에 '반월식당'을 비롯한 글자가 적힌 간판 및 메뉴판이 있습니다.",
     "fix_en": "Remove the sign with Korean text and the menu board below it, replacing them with bare wood matching the surrounding interior structure.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the sign with Korean text and the menu board below it, replacing them with bare wood matching the surrounding interior structure.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트+seed만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_bypass:bg_only"
 },
 "S90sh4": {
  "input_fingerprint": "a836712f580e1aac",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk, reddish sunset.\n\nSHOT TEXT (authoritative, Korean): 아무도 없는 빈 홀 한가운데서 의아한 표정으로 멈춰 선 김지후(강수리영)의 전신.\n\nLOCATION (lock): In the middle of the deserted restaurant dining hall under the fading sunset glow. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the empty restaurant hall, reddish sunset light, tables, kitchen, and silent atmosphere. Exclude the purely empty establishing composition and add the woman searching alone.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 5,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "빈 홀 한가운데 멈춰 선 전신 숏과 의아해하는 표정을 가장 자연스럽고 충실하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "전신 숏과 정지한 자세는 지켰으나, 의아한 표정 대신 레퍼런스 이미지와 동일한 무표정으로 렌더링되었습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "표정은 나쁘지 않으나, 발을 내딛으며 걷는 자세로 묘사되어 '멈춰 선'이라는 핵심 행동 지시를 어겼습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S90sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 샷의 레퍼런스와 동일한 카메라 구도를 사용하지 말라는 지시를 어기고 완전히 똑같은 앵글과 구도로 렌더링되었습니다.",
     "fix_en": "The camera framing exactly copies the previous shot; this requires moving the camera to a new angle.",
     "unfixable": true
    },
    {
     "issue_ko": "지문에서 요구한 '의아한 표정'이 나타나지 않고 무표정한 얼굴을 하고 있습니다.",
     "fix_en": "Modify her facial expression to look puzzled and questioning by slightly furrowing her eyebrows.",
     "unfixable": false
    },
    {
     "issue_ko": "여성의 오른쪽 발이 기존 배경 이미지 바닥에 있던 검은색 슬리퍼와 부자연스럽게 합쳐져 있습니다.",
     "fix_en": "Redraw her right shoe to clearly stand on the floor, removing the dark slippers from the background that overlap it.",
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Modify her facial expression to look puzzled and questioning by slightly furrowing her eyebrows.\n- Redraw her right shoe to clearly stand on the floor, removing the dark slippers from the background that overlap it.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S90sh1"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "groupbg::식당_뒤뜰": {
  "input_fingerprint": "931d17db20640d7b",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "식당_뒤뜰",
    "tags": [
     "S26sh3",
     "S26sh4",
     "S28sh2",
     "S37sh10",
     "S37sh5",
     "S66sh1",
     "S72sh2",
     "S72sh4",
     "S91sh1"
    ]
   },
   "context_sig": "b6ee62cafc0b1aee"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: In the restaurant's rear yard at sunset, between the building and the burned storage area.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 갯바위: 거친 파도가 철썩이며 부딪히는 해안가의 검고 날카로운 바위 지대. (특징: 검고 날카로운 해안 암석들; 하얗게 부서지는 거친 파도와 포말; 미끄럽게 젖은 바위 표면) / 금월도 숲속 동굴 내부: 빛이 들지 않는 깊고 축축한 자연 동굴. 바닥에 인골과 토막 난 시체가 널려 있다. (특징: 울퉁불퉁하고 어두운 암석 벽과 천장; 천장에 빽빽이 매달린 수백 마리 박쥐 떼; 바닥에 흩어진 사람의 뼈와 토막 난 신체 부위; 바닥에 고인 피 웅덩이; 벽에 고정된 무거운 쇠사슬) / 금월도 숲속 동굴 입구: 거대한 자연 동굴의 입구. 나무 기둥과 판자를 얼기설기 덧대어 형태를 만들었고 내부는 캄캄하다. (특징: 거대한 암석 동굴 구멍; 입구에 덧댄 낡은 나무 기둥과 판자; 캄캄하고 어두운 동굴 안쪽) / 반월 식당 뒤뜰 창고 내부: 식당 뒤뜰에 있는 허름한 창고의 내부. 밤에는 기괴한 제단이 켜지고 낮에는 낚시 장비가 쌓인 모습이다. (특징: 얼기설기한 나무 벽면; 수십 개의 촛불과 향이 피어오르는 무속 제단; 제단 위 검은 염소 머리와 피가 담긴 사발; 사발 위에 놓인 하얀 사슴뿔 창; 잡다한 낚시 용품 및 비품 더미) / 반월 식당 뒷방: 식당 건물 안쪽에 위치한 주거용 방. 정갈하게 정돈되어 햇살이 스며든다. (특징: 정돈된 방바닥과 이부자리; 격자무늬 나무 창; 창문 너머 들어오는 햇살; 옷걸이에 걸린 화려한 붉은색 무녀복) / 반월 식당 앞: 식당 건물 앞의 골목길 및 입구 흙길 지형. (특징: 미닫이 나무 문; 낡은 식당 외벽; 흙 덮인 골목길; 바닥의 짙은 핏자국)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 반월 식당 뒤뜰 - 밤\n- 잘 꾸민 텃밭과 평상 위에 말려둔 시래기 같은 채소 뿐…\n- 실외. 반월 식당 뒤뜰 - 낮\n- 실외. 반월 식당 뒤뜰 - 밤\n- 실외. 반월 식당 뒤뜰 - 밤\n- 실외. 식당 뒤뜰 - 해질녘\n\nTIME OF DAY (lock): dusk.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: In the restaurant's rear yard at sunset, between the building and the burned storage area.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 갯바위: 거친 파도가 철썩이며 부딪히는 해안가의 검고 날카로운 바위 지대. (특징: 검고 날카로운 해안 암석들; 하얗게 부서지는 거친 파도와 포말; 미끄럽게 젖은 바위 표면) / 금월도 숲속 동굴 내부: 빛이 들지 않는 깊고 축축한 자연 동굴. 바닥에 인골과 토막 난 시체가 널려 있다. (특징: 울퉁불퉁하고 어두운 암석 벽과 천장; 천장에 빽빽이 매달린 수백 마리 박쥐 떼; 바닥에 흩어진 사람의 뼈와 토막 난 신체 부위; 바닥에 고인 피 웅덩이; 벽에 고정된 무거운 쇠사슬) / 금월도 숲속 동굴 입구: 거대한 자연 동굴의 입구. 나무 기둥과 판자를 얼기설기 덧대어 형태를 만들었고 내부는 캄캄하다. (특징: 거대한 암석 동굴 구멍; 입구에 덧댄 낡은 나무 기둥과 판자; 캄캄하고 어두운 동굴 안쪽) / 반월 식당 뒤뜰 창고 내부: 식당 뒤뜰에 있는 허름한 창고의 내부. 밤에는 기괴한 제단이 켜지고 낮에는 낚시 장비가 쌓인 모습이다. (특징: 얼기설기한 나무 벽면; 수십 개의 촛불과 향이 피어오르는 무속 제단; 제단 위 검은 염소 머리와 피가 담긴 사발; 사발 위에 놓인 하얀 사슴뿔 창; 잡다한 낚시 용품 및 비품 더미) / 반월 식당 뒷방: 식당 건물 안쪽에 위치한 주거용 방. 정갈하게 정돈되어 햇살이 스며든다. (특징: 정돈된 방바닥과 이부자리; 격자무늬 나무 창; 창문 너머 들어오는 햇살; 옷걸이에 걸린 화려한 붉은색 무녀복) / 반월 식당 앞: 식당 건물 앞의 골목길 및 입구 흙길 지형. (특징: 미닫이 나무 문; 낡은 식당 외벽; 흙 덮인 골목길; 바닥의 짙은 핏자국)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 반월 식당 뒤뜰 - 밤\n- 잘 꾸민 텃밭과 평상 위에 말려둔 시래기 같은 채소 뿐…\n- 실외. 반월 식당 뒤뜰 - 낮\n- 실외. 반월 식당 뒤뜰 - 밤\n- 실외. 반월 식당 뒤뜰 - 밤\n- 실외. 식당 뒤뜰 - 해질녘\n\nTIME OF DAY (lock): dusk.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_식당_뒤뜰_67967e.png",
  "asset_id": "10c72334-0e67-4d2c-b412-b0cb37a597db",
  "input_asset_ids": [
   "2a1a4d1c-8dbf-4632-a037-6b2090bd486e"
  ],
  "origin_tag": "S91sh1",
  "place_text": "In the restaurant's rear yard at sunset, between the building and the burned storage area.",
  "origin_inputs": {
   "place_text": "In the restaurant's rear yard at sunset, between the building and the burned storage area.",
   "time_of_day_en": "dusk",
   "conti_asset_id": "2a1a4d1c-8dbf-4632-a037-6b2090bd486e"
  }
 },
 "S91sh1::bgfirst_bg": {
  "input_fingerprint": "11d4fbf8606af895",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 식당 뒤뜰에 서서 고개를 옆으로 반쯤 돌린 채 불안한 눈빛으로 주변을 응시하는 김지후(강수리영).\n\nLOCATION (lock): In the restaurant's rear yard at sunset, between the building and the burned storage area.\n\nTIME OF DAY (lock): dusk.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height at a three-quarter frontal angle, observe 김지후(강수리영) at medium distance as her weight shifts during a cautious scan of the backyard. Hold her in one side of the frame and preserve open space in the direction of her half-turned head, preparing the straight push-in that follows the same gaze axis.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 식당 뒤뜰 (강민숙(홍은미) is not visible anywhere in the shown area); used as Surrounding negative space through which her searching gaze moves.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 식당 뒤뜰에 서서 고개를 옆으로 반쯤 돌린 채 불안한 눈빛으로 주변을 응시하는 김지후(강수리영).\n\nLOCATION (lock): In the restaurant's rear yard at sunset, between the building and the burned storage area.\n\nTIME OF DAY (lock): dusk.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height at a three-quarter frontal angle, observe 김지후(강수리영) at medium distance as her weight shifts during a cautious scan of the backyard. Hold her in one side of the frame and preserve open space in the direction of her half-turned head, preparing the straight push-in that follows the same gaze axis.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 식당 뒤뜰 (강민숙(홍은미) is not visible anywhere in the shown area); used as Surrounding negative space through which her searching gaze moves.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S91sh1__bgfirst_bg.png",
  "asset_id": "ec23e78f-7cd1-4824-ae97-fc897bcaccae",
  "input_asset_ids": [
   "2a1a4d1c-8dbf-4632-a037-6b2090bd486e",
   "10c72334-0e67-4d2c-b412-b0cb37a597db"
  ]
 },
 "S91sh1": {
  "input_fingerprint": "38718cfc0933a8fa",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 식당 뒤뜰에 서서 고개를 옆으로 반쯤 돌린 채 불안한 눈빛으로 주변을 응시하는 김지후(강수리영).\n\nLOCATION (lock): In the restaurant's rear yard at sunset, between the building and the burned storage area. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 식당 뒤뜰에 서서 고개를 옆으로 반쯤 돌린 채 불안한 눈빛으로 주변을 응시하는 김지후(강수리영).\n\nLOCATION (lock): In the restaurant's rear yard at sunset, between the building and the burned storage area. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 식당 뒤뜰에 서서 고개를 옆으로 반쯤 돌린 채 불안한 눈빛으로 주변을 응시하는 김지후(강수리영).\n\nLOCATION (lock): In the restaurant's rear yard at sunset, between the building and the burned storage area. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S91sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S91sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_식당_뒤뜰_67967e.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 식당 뒤뜰을 정확히 반영하며, 고개를 반쯤 돌린 채 불안한 눈빛을 띠는 인물의 상태를 적절한 프레이밍으로 잘 담아냈습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "요구된 인물의 자세와 시선 처리는 구현되었으나, 원본 레퍼런스에 존재하지 않는 초록색 플라스틱 상자가 임의로 배경에 추가되었습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 식당 뒤뜰을 정확히 반영하며, 고개를 반쯤 돌린 채 불안한 눈빛을 띠는 인물의 상태를 적절한 프레이밍으로 잘 담아냈습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "요구된 인물의 자세와 시선 처리는 구현되었으나, 원본 레퍼런스에 존재하지 않는 초록색 플라스틱 상자가 임의로 배경에 추가되었습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 로케이션을 정확히 구현하였으며, 인물의 불안한 눈빛과 반쯤 돌린 고개를 적절한 프레이밍으로 잘 표현했습니다."
     },
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "로케이션과 인물의 자세는 지문에 부합하나, 프레임이 넓어 '불안한 눈빛'의 감정 표현이 덜 돋보이며 바닥에 레퍼런스에 없는 상자가 추가되었습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지정된 로케이션을 정확히 구현하였으며, 인물의 불안한 눈빛과 반쯤 돌린 고개를 적절한 프레이밍으로 잘 표현했습니다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "로케이션과 인물의 자세는 지문에 부합하나, 프레임이 넓어 '불안한 눈빛'의 감정 표현이 덜 돋보이며 바닥에 레퍼런스에 없는 상자가 추가되었습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 12
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 15,
   "B": 12
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지정된 식당 뒤뜰을 정확히 반영하며, 고개를 반쯤 돌린 채 불안한 눈빛을 띠는 인물의 상태를 적절한 프레이밍으로 잘 담아냈습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "요구된 인물의 자세와 시선 처리는 구현되었으나, 원본 레퍼런스에 존재하지 않는 초록색 플라스틱 상자가 임의로 배경에 추가되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_식당_뒤뜰_67967e.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "물건을 쥐고 있는 양손의 손가락 형태가 서로 뭉개져 있어 해부학적으로 어색합니다.",
     "fix_en": "Correct the anatomy of the fingers on both hands so they are naturally defined while holding the object.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Correct the anatomy of the fingers on both hands so they are naturally defined while holding the object.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S91sh1__bgfirst_bg.png",
   "bg_asset_id": "ec23e78f-7cd1-4824-ae97-fc897bcaccae",
   "bg_record_key": "S91sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "식당_뒤뜰",
   "groupbg_asset_id": "10c72334-0e67-4d2c-b412-b0cb37a597db"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S92sh3::bgfirst_bg": {
  "input_fingerprint": "40d4379f640e2a6c",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 노을 지는 선착장 끝 돌바닥에 앉아 바다를 향해 있는 강민숙(홍은미)의 뚜렷한 뒷모습.\n\nLOCATION (lock): Directly outside the restaurant's front entrance at sunset, looking toward the island pier.\n\nTIME OF DAY (lock): dusk.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Centered behind 강민숙(홍은미) at the supplied low seated-axis height, continue the dolly-in until her clearly defined back and seated body anchor the lower middle of a wide frame. The stone dock leads past her toward the sea, while her unmoving orientation keeps her face completely withheld.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: sea beyond the dock in the upper-center of the frame, background; end of the stone dock in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선착장 돌바닥 (강민숙(홍은미) is seated at its far end); used as Leading surface carrying the eye from the camera to the seated figure; 바다 (Visible beyond the end of the dock at sunset); used as The distant field directly ahead of the seated figure.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 노을 지는 선착장 끝 돌바닥에 앉아 바다를 향해 있는 강민숙(홍은미)의 뚜렷한 뒷모습.\n\nLOCATION (lock): Directly outside the restaurant's front entrance at sunset, looking toward the island pier.\n\nTIME OF DAY (lock): dusk.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Centered behind 강민숙(홍은미) at the supplied low seated-axis height, continue the dolly-in until her clearly defined back and seated body anchor the lower middle of a wide frame. The stone dock leads past her toward the sea, while her unmoving orientation keeps her face completely withheld.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: sea beyond the dock in the upper-center of the frame, background; end of the stone dock in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선착장 돌바닥 (강민숙(홍은미) is seated at its far end); used as Leading surface carrying the eye from the camera to the seated figure; 바다 (Visible beyond the end of the dock at sunset); used as The distant field directly ahead of the seated figure.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S92sh3__bgfirst_bg.png",
  "asset_id": "9a151075-adf5-4321-9ace-fcdb4c35b73c",
  "input_asset_ids": [
   "2385d556-f2ad-4fd4-981d-8edab32ad2ae",
   "7635b014-5030-4fd4-bbf7-e864105269ed"
  ]
 },
 "S92sh3": {
  "input_fingerprint": "999e1470fda255ee",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 노을 지는 선착장 끝 돌바닥에 앉아 바다를 향해 있는 강민숙(홍은미)의 뚜렷한 뒷모습.\n\nLOCATION (lock): Directly outside the restaurant's front entrance at sunset, looking toward the island pier. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 노을 지는 선착장 끝 돌바닥에 앉아 바다를 향해 있는 강민숙(홍은미)의 뚜렷한 뒷모습.\n\nLOCATION (lock): Directly outside the restaurant's front entrance at sunset, looking toward the island pier. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 노을 지는 선착장 끝 돌바닥에 앉아 바다를 향해 있는 강민숙(홍은미)의 뚜렷한 뒷모습.\n\nLOCATION (lock): Directly outside the restaurant's front entrance at sunset, looking toward the island pier. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S92sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S92sh3.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L18B02.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "선착장 돌바닥에 앉아 바다를 향한 뚜렷한 뒷모습을 잘 구현했으며, 금지된 레퍼런스의 카메라 구도를 피했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "돌바닥이 아닌 나무 구조물 위에 앉아 샷 텍스트를 위반했으며, 금지된 레퍼런스의 카메라 구도(양옆 프레임)를 그대로 복사했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "선착장 돌바닥에 앉아 바다를 향한 뚜렷한 뒷모습을 잘 구현했으며, 금지된 레퍼런스의 카메라 구도를 피했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "돌바닥이 아닌 나무 구조물 위에 앉아 샷 텍스트를 위반했으며, 금지된 레퍼런스의 카메라 구도(양옆 프레임)를 그대로 복사했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "선착장 돌바닥에 앉아 바다를 응시하는 뚜렷한 뒷모습을 지문대로 정확하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "지문에서 요구한 돌바닥이 아닌 나무 구조물 위에 앉아 있으며, 완전한 뒷모습이 아닌 측후면이 보입니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "선착장 돌바닥에 앉아 바다를 응시하는 뚜렷한 뒷모습을 지문대로 정확하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "지문에서 요구한 돌바닥이 아닌 나무 구조물 위에 앉아 있으며, 완전한 뒷모습이 아닌 측후면이 보입니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 7
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 7
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "선착장 돌바닥에 앉아 바다를 향한 뚜렷한 뒷모습을 잘 구현했으며, 금지된 레퍼런스의 카메라 구도를 피했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "돌바닥이 아닌 나무 구조물 위에 앉아 샷 텍스트를 위반했으며, 금지된 레퍼런스의 카메라 구도(양옆 프레임)를 그대로 복사했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L18B02.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "인물이 레이아웃 스케치에 지정된 위치보다 너무 크고 카메라에 가깝게 배치되었습니다.",
     "fix_en": "Scale down the woman and move her further down the pier to match the exact size and position in the layout sketch.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "스케치의 화살표가 오른쪽을 향하고 있으나, 인물이 정면을 바라보고 있습니다.",
     "fix_en": "Turn the woman's head to look toward the right, following the sketch's arrow.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Scale down the woman and move her further down the pier to match the exact size and position in the layout sketch.\n- Turn the woman's head to look toward the right, following the sketch's arrow.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S92sh3__bgfirst_bg.png",
   "bg_asset_id": "9a151075-adf5-4321-9ace-fcdb4c35b73c",
   "bg_record_key": "S92sh3::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S93sh5": {
  "input_fingerprint": "f58b7f3b0cb0d2e2",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk, warm sunset.\n\nSHOT TEXT (authoritative, Korean): 어깨 너머로 고개를 돌린 채 섬뜩한 미소를 짓고 있는 강영란(흰 머리 여자)의 찰나.\n\nLOCATION (lock): At the far end of the stone pier at sunset, facing the gold-lit sea. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "B": 5,
   "A": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "엄격한 인물 제한(김지후)과 이전 샷의 의상을 완벽히 유지했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "이전 샷의 의상은 유지했으나 인물 제한 규칙(김지후)을 위반했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "인물 제한 규칙을 위반하고 이전 샷의 의상까지 변경했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S92sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1225321>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "샷 텍스트는 '섬뜩한 미소'를 짓고 있다고 명시했으나, 현재 이미지 속 인물은 밝고 다정한 미소를 짓고 있습니다.",
     "fix_en": "Change the woman's expression to an eerie, unsettling smile.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "샷 텍스트에서 '흰 머리 여자'를 명시적으로 요구하고 있으나, 인물의 머리카락이 검은색으로 표현되었습니다.",
     "fix_en": "Change the woman's hair color from black to white.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the woman's expression to an eerie, unsettling smile.\n- Change the woman's hair color from black to white.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S92sh3"
  }
 },
 "S93sh8": {
  "input_fingerprint": "29505a920e490f8d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk, warm sunset.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)의 어깨에서 벗어나 바닥을 향해 반쯤 떨어져 내리며 허공에 떠 있는 창백한 강민숙(홍은미)의 젖은 시체.\n\nLOCATION (lock): At the sunset pier's seaward end, where the seated figure holds a soaked body. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the sunset pier, golden sea, seated elder, and shoreline light. Exclude her earlier concealed burden and reveal the wet corpse slipping from her shoulder.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): Her soaked corpse hangs completely limp in Youngran's arms, with her body supported against Youngran and her pale head dropping downward from behind Youngran's shoulder.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Within Jihu’s vision, Minsuk’s soaked corpse hangs completely limp in Youngran's arms, with her body supported against Youngran and her pale head dropping downward from behind Youngran's shoulder.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 3,
   "B": 7,
   "C": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "시체가 어깨 뒤로 매달려 있지 않고 바닥에 앉아 있어 중력 및 캐논 자세 지침을 완전히 위반했습니다."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "어깨 뒤에서 반쯤 떨어져 허공에 늘어진 시체의 캐논 자세와 지문의 묘사를 가장 충실하게 구현했습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "시체가 어깨 뒤가 아닌 앞쪽에서 안겨 있는 형태로 묘사되어 샷 텍스트의 핵심적인 자세를 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S93sh5_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1225321>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "시체의 몸통이 중력을 무시하고 허공에 수평으로 떠 있으며, 영란에게 지탱되어 축 늘어진 자세라는 지시를 위반했습니다.",
     "fix_en": "Redraw the corpse's torso to hang vertically and limply straight down against the woman's back, obeying gravity.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "영란의 무릎 위에 신체와 연결되지 않은 정체불명의 창백한 손과 어두운 색 소매가 놓여 있습니다.",
     "fix_en": "Remove the disembodied hand and dark sleeve resting on the woman's lap, filling the space with the dark trousers.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the corpse's torso to hang vertically and limply straight down against the woman's back, obeying gravity.\n- Remove the disembodied hand and dark sleeve resting on the woman's lap, filling the space with the dark trousers.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S93sh5"
  }
 },
 "S93sh10": {
  "input_fingerprint": "07913b87879d4781",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): dusk, warm sunset.\n\nSHOT TEXT (authoritative, Korean): 극도의 공포에 질려 눈물이 고인 채 잔뜩 경직되어 서 있는 김지후(강수리영)의 전신.\n\nLOCATION (lock): On the stone pier a few steps from its seaward end and the seated pair. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the golden harbor, pier stones, seated elder, and wet corpse. Exclude the elder-centered framing and show the younger woman frozen in horror before them.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Within the same vision, Minsuk’s corpse remains limp and wet in Youngran’s embrace.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 5,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지시된 '전신' 프레이밍을 정확히 구현했으며, 배경 인물들의 고정된 자세와 안착 상태를 허용 가능한 수준으로 잘 유지했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "배경 인물의 자세와 배치는 가장 완벽하지만, 텍스트가 요구한 '전신' 숏이 아닌 미디엄 숏으로 렌더링되어 프레임 우선순위에서 감점되었습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "전신 프레임은 충족했으나, 원본과 동일하게 유지되어야 할 시체의 자세가 완전히 엎드린 형태로 왜곡되어 심각한 오류가 발생했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S93sh8_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "기준 이미지와 달리 시체의 팔 자세가 변형되었습니다. 시체 얼굴 옆으로 보이지 않던 팔이 늘어져 있고, 노인 등 뒤로 곧게 늘어져 있어야 할 시체의 반대쪽 팔은 밖을 향해 구부러져 있습니다.",
     "fix_en": "Remove the corpse's arm dangling near its face, and straighten its other arm behind the elder's back so it hangs straight down.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the corpse's arm dangling near its face, and straighten its other arm behind the elder's back so it hangs straight down.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S93sh8"
  }
 },
 "S94sh2::bgfirst_bg": {
  "input_fingerprint": "dbe5cb00ce2e1a09",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 맞은편 동굴 벽에 쇠사슬로 묶인 채 얼굴이 상처투성이가 된 정인우의 전신.\n\nLOCATION (lock): Inside the cave prison in daylight, at the opposite wall where a battered captive is chained.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From mid-height near 김지후(강수리영)'s side of the cave, finish the pan on a wide full-body view of 정인우 against the opposite wall. Hold him slightly off-center with his chained limbs and damaged face visible in one continuous silhouette, preserving the empty distance across the cave as the axis back toward 김지후(강수리영).\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 맞은편 동굴 벽 (정인우 is restrained against it); used as The backing plane that fixes 정인우's restrained full-body position; 쇠사슬 (Fastened tightly around 정인우 and securing him to the wall); used as Visible restraint connecting his body to the wall; 동굴 내부 공간 (Separates the two restrained characters on opposite sides); used as Open spatial interval that establishes separation from 김지후(강수리영)'s side.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 맞은편 동굴 벽에 쇠사슬로 묶인 채 얼굴이 상처투성이가 된 정인우의 전신.\n\nLOCATION (lock): Inside the cave prison in daylight, at the opposite wall where a battered captive is chained.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From mid-height near 김지후(강수리영)'s side of the cave, finish the pan on a wide full-body view of 정인우 against the opposite wall. Hold him slightly off-center with his chained limbs and damaged face visible in one continuous silhouette, preserving the empty distance across the cave as the axis back toward 김지후(강수리영).\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 맞은편 동굴 벽 (정인우 is restrained against it); used as The backing plane that fixes 정인우's restrained full-body position; 쇠사슬 (Fastened tightly around 정인우 and securing him to the wall); used as Visible restraint connecting his body to the wall; 동굴 내부 공간 (Separates the two restrained characters on opposite sides); used as Open spatial interval that establishes separation from 김지후(강수리영)'s side.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S94sh2__bgfirst_bg.png",
  "asset_id": "d0603283-5aeb-4d72-9061-a299ee934cb2",
  "input_asset_ids": [
   "f9ebc7be-ddef-4f89-b7b1-e70d630eaf2f",
   "7fdfffd1-1496-45ab-9347-250f4acbf701"
  ]
 },
 "S94sh2": {
  "input_fingerprint": "120535f2f760511f",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 맞은편 동굴 벽에 쇠사슬로 묶인 채 얼굴이 상처투성이가 된 정인우의 전신.\n\nLOCATION (lock): Inside the cave prison in daylight, at the opposite wall where a battered captive is chained. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): He is restrained upright against the opposite cave wall by heavy chains, with his back held close to the rock and his arms and legs prevented from carrying him away.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo remains restrained upright against the opposite cave wall by heavy chains, with his back held close to the rock and his arms and legs prevented from carrying him away, his face scratched and split; Jihu is likewise securely chained.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 맞은편 동굴 벽에 쇠사슬로 묶인 채 얼굴이 상처투성이가 된 정인우의 전신.\n\nLOCATION (lock): Inside the cave prison in daylight, at the opposite wall where a battered captive is chained. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): He is restrained upright against the opposite cave wall by heavy chains, with his back held close to the rock and his arms and legs prevented from carrying him away.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo remains restrained upright against the opposite cave wall by heavy chains, with his back held close to the rock and his arms and legs prevented from carrying him away, his face scratched and split; Jihu is likewise securely chained.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 맞은편 동굴 벽에 쇠사슬로 묶인 채 얼굴이 상처투성이가 된 정인우의 전신.\n\nLOCATION (lock): Inside the cave prison in daylight, at the opposite wall where a battered captive is chained. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): He is restrained upright against the opposite cave wall by heavy chains, with his back held close to the rock and his arms and legs prevented from carrying him away.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo remains restrained upright against the opposite cave wall by heavy chains, with his back held close to the rock and his arms and legs prevented from carrying him away, his face scratched and split; Jihu is likewise securely chained.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S94sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S94sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1225321>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L53B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1225321>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 로케이션 레퍼런스의 건축 구조(콘크리트 벽, 문)를 정확히 유지하고 반대편 벽에 인물을 배치한 전신 샷을 잘 구현했으나, 얼굴의 상처 묘사는 다소 약합니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "얼굴의 상처와 복장은 잘 표현되었으나, 로케이션의 오른쪽 벽과 배경을 임의로 자연 암벽 동굴로 변형시켜 레퍼런스의 건축 구조를 심각하게 훼손했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 로케이션 레퍼런스의 건축 구조(콘크리트 벽, 문)를 정확히 유지하고 반대편 벽에 인물을 배치한 전신 샷을 잘 구현했으나, 얼굴의 상처 묘사는 다소 약합니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "얼굴의 상처와 복장은 잘 표현되었으나, 로케이션의 오른쪽 벽과 배경을 임의로 자연 암벽 동굴로 변형시켜 레퍼런스의 건축 구조를 심각하게 훼손했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "로케이션 레퍼런스의 건축 양식(매끄러운 벽, 열린 문)을 정확히 유지하면서, 맞은편 벽에 결박된 인물의 전신을 지시문대로 잘 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "로케이션 레퍼런스의 매끄러운 벽과 문을 무시하고 공간을 자연 암석 동굴로 임의로 변경하여 위치 제약을 심각하게 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "로케이션 레퍼런스의 건축 양식(매끄러운 벽, 열린 문)을 정확히 유지하면서, 맞은편 벽에 결박된 인물의 전신을 지시문대로 잘 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "로케이션 레퍼런스의 매끄러운 벽과 문을 무시하고 공간을 자연 암석 동굴로 임의로 변경하여 위치 제약을 심각하게 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 6
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지정된 로케이션 레퍼런스의 건축 구조(콘크리트 벽, 문)를 정확히 유지하고 반대편 벽에 인물을 배치한 전신 샷을 잘 구현했으나, 얼굴의 상처 묘사는 다소 약합니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "얼굴의 상처와 복장은 잘 표현되었으나, 로케이션의 오른쪽 벽과 배경을 임의로 자연 암벽 동굴로 변형시켜 레퍼런스의 건축 구조를 심각하게 훼손했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L53B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1225321>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트 규정상 김지후(20대 여성)만 등장할 수 있으나, 샷 텍스트의 이름 때문에 남성이 묘사되었습니다.",
     "fix_en": "Replace the male character with the young woman Kim Ji-hu from the character reference, maintaining the exact same chained pose and position.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "샷 텍스트와 상태 설명에 명시된 '얼굴이 상처투성이가 된' 모습이 뚜렷하게 표현되지 않았습니다.",
     "fix_en": "Add severe cuts, scratches, and wounds all over the character's face.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "인물의 양쪽 손목 수갑에서 벽으로 이어지는 쇠사슬 외에, 아래로 길게 늘어지는 불필요한 쇠사슬 가닥이 추가로 생성되었습니다.",
     "fix_en": "Erase the extra lengths of chain hanging downwards from both wrist manacles, leaving only the taut chains connecting the wrists to the wall plates.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the male character with the young woman Kim Ji-hu from the character reference, maintaining the exact same chained pose and position.\n- Add severe cuts, scratches, and wounds all over the character's face.\n- Erase the extra lengths of chain hanging downwards from both wrist manacles, leaving only the taut chains connecting the wrists to the wall plates.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S94sh2__bgfirst_bg.png",
   "bg_asset_id": "d0603283-5aeb-4d72-9061-a299ee934cb2",
   "bg_record_key": "S94sh2::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S94sh7": {
  "input_fingerprint": "602b12ee937c2752",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 삶의 의지를 모두 잃은 듯 텅 빈 눈동자로 바닥을 멍하니 응시하는 김지후(강수리영)의 모습.\n\nLOCATION (lock): On the cave floor at the chained holding area, dim despite the daytime outside. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the rough cave walls, chains, dim daylight, and the injured man bound opposite. Exclude his full-body emphasis and show the woman staring blankly at the floor.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is seated in a defeated slump on the cave floor with her body and limbs secured by heavy chains, her head bowed and her gaze lowered toward the ground.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu is seated in a defeated slump on the cave floor with her body and limbs secured by heavy chains, her head bowed and her gaze lowered toward the ground; Inwoo remains securely chained to the opposite cave wall and retains his facial injuries.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 2
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "이전 샷의 벽면 재질과 고정 장치를 정확히 유지하고 우측 배경에 결박된 남자의 서 있는 자세(3순위)를 완벽히 지켜냈으나, 여성 캐릭터의 의상이 섞여 나타난 점(4순위)이 아쉽습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "여성 캐릭터의 의상(4순위)은 잘 구현되었으나, 이전 샷의 미장 벽면 대신 자연 암석을 배경으로 삼고 남자를 주저앉혀 장소 및 자세 유지 지침(3순위)을 어겼습니다."
   },
   {
    "label": "C",
    "score": 2,
    "verdict_ko": "비유적 표현을 문자 그대로 받아들여 눈동자를 하얗게 지운 것은 해부학적 변형 금지 지침을 어긴 심각한 위반(1순위)입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S94sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "김지후가 캐릭터 레퍼런스에 지정된 의상(어두운 스웨터와 청바지)이 아닌, 이전 컷의 인물(인우)이 입고 있던 회색 셔츠와 검은색 바지를 입고 있습니다.",
     "fix_en": "Change Kim Ji-hoo's clothing to a dark cable-knit sweater and blue jeans to match her character reference."
    },
    {
     "issue_ko": "화면 오른쪽 끝에 보이는 인우의 발이 벽 쪽(오른쪽)을 향하고 있습니다. 이전 컷에서 벽을 등지고 있었으므로 발끝이 바깥쪽(왼쪽)을 향해야 합니다.",
     "fix_en": "Redraw the foot on the far right edge so that the heel is near the wall and the toes point outwards to the left."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change Kim Ji-hoo's clothing to a dark cable-knit sweater and blue jeans to match her character reference.\n- Redraw the foot on the far right edge so that the heel is near the wall and the toes point outwards to the left.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S94sh2"
  }
 },
 "groupbg::해안절벽_정상": {
  "input_fingerprint": "0561a83048196f78",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "해안절벽_정상",
    "tags": [
     "S95sh1",
     "S95sh10",
     "S95sh18"
    ]
   },
   "context_sig": "df4457de17fcccd7"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: At the exposed edge of a high island sea cliff under a bright sky.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 해안 절벽 - 낮\n- 붉은색 무녀복 차림의 영란과 황 씨가 절벽에 서서 바다를 바라본다.\n\nTIME OF DAY (lock): day, clear sky.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: At the exposed edge of a high island sea cliff under a bright sky.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 선착장: 섬으로 들어오는 시멘트 바닥의 선착장. 밧줄 묶는 쇠기둥과 어선들이 대어 있고 주변에 숲이 보인다. (특징: 시멘트 바닥; 배를 묶는 선착장 기둥; 주변에 정박한 어선들; 사건 현장의 노란색 폴리스 라인)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 해안 절벽 - 낮\n- 붉은색 무녀복 차림의 영란과 황 씨가 절벽에 서서 바다를 바라본다.\n\nTIME OF DAY (lock): day, clear sky.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해안절벽_정상_8b9a7f.png",
  "asset_id": "61c6f3c2-be55-452d-853a-13950342d6d5",
  "input_asset_ids": [
   "6a40399a-8f33-4d83-8dfb-8c3feea8ebd9"
  ],
  "origin_tag": "S95sh1",
  "place_text": "At the exposed edge of a high island sea cliff under a bright sky.",
  "origin_inputs": {
   "place_text": "At the exposed edge of a high island sea cliff under a bright sky.",
   "time_of_day_en": "day, clear sky",
   "conti_asset_id": "6a40399a-8f33-4d83-8dfb-8c3feea8ebd9"
  }
 },
 "S95sh1::bgfirst_bg": {
  "input_fingerprint": "e8940f4d956108c8",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 눈부신 하늘과 바다를 배경으로 절벽 끝에 선 강영란(흰 머리 여자)의 옷자락이 바람에 흩날리는 전경.\n\nLOCATION (lock): At the exposed edge of a high island sea cliff under a bright sky.\n\nTIME OF DAY (lock): day, clear sky.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the inland side, hold a static eye-level wide frame facing out toward the sea, with 강영란(흰 머리 여자) appearing as a small full figure near the cliff edge and the sky and sea occupying most of the image. Her back and wind-lifted red skirt create the only strong vertical and moving accents within the broad horizontal landscape.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: cliff edge (강영란(흰 머리 여자) is positioned at its end) — The inland top and abrupt seaward edge are visible in profile across the lower frame; used as Forms the narrow ground line beneath the isolated figure; daytime sky (Bright and blue); used as Occupies a large portion of the background behind the figure; sea (Blue and visibly rolling); used as Creates the outward vista and distant visual destination of her gaze.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 눈부신 하늘과 바다를 배경으로 절벽 끝에 선 강영란(흰 머리 여자)의 옷자락이 바람에 흩날리는 전경.\n\nLOCATION (lock): At the exposed edge of a high island sea cliff under a bright sky.\n\nTIME OF DAY (lock): day, clear sky.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the inland side, hold a static eye-level wide frame facing out toward the sea, with 강영란(흰 머리 여자) appearing as a small full figure near the cliff edge and the sky and sea occupying most of the image. Her back and wind-lifted red skirt create the only strong vertical and moving accents within the broad horizontal landscape.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: cliff edge (강영란(흰 머리 여자) is positioned at its end) — The inland top and abrupt seaward edge are visible in profile across the lower frame; used as Forms the narrow ground line beneath the isolated figure; daytime sky (Bright and blue); used as Occupies a large portion of the background behind the figure; sea (Blue and visibly rolling); used as Creates the outward vista and distant visual destination of her gaze.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S95sh1__bgfirst_bg.png",
  "asset_id": "c9ca11ca-5b62-4c6d-9b2f-e82406a3b54f",
  "input_asset_ids": [
   "6a40399a-8f33-4d83-8dfb-8c3feea8ebd9",
   "61c6f3c2-be55-452d-853a-13950342d6d5"
  ]
 },
 "S95sh1": {
  "input_fingerprint": "eac4f7c9e3258447",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, clear sky.\n\nSHOT TEXT (authoritative, Korean): 눈부신 하늘과 바다를 배경으로 절벽 끝에 선 강영란(흰 머리 여자)의 옷자락이 바람에 흩날리는 전경.\n\nLOCATION (lock): At the exposed edge of a high island sea cliff under a bright sky. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran remains dressed in the red shaman’s robe; Sandeul retains the wounds from the patrol-boat struggle.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, clear sky.\n\nSHOT TEXT (authoritative, Korean): 눈부신 하늘과 바다를 배경으로 절벽 끝에 선 강영란(흰 머리 여자)의 옷자락이 바람에 흩날리는 전경.\n\nLOCATION (lock): At the exposed edge of a high island sea cliff under a bright sky. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran remains dressed in the red shaman’s robe; Sandeul retains the wounds from the patrol-boat struggle.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, clear sky.\n\nSHOT TEXT (authoritative, Korean): 눈부신 하늘과 바다를 배경으로 절벽 끝에 선 강영란(흰 머리 여자)의 옷자락이 바람에 흩날리는 전경.\n\nLOCATION (lock): At the exposed edge of a high island sea cliff under a bright sky. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran remains dressed in the red shaman’s robe; Sandeul retains the wounds from the patrol-boat struggle.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S95sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S95sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1010229>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해안절벽_정상_8b9a7f.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1010229>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 로케이션과 앵글을 잘 유지했으며, 레퍼런스의 붉은 무당복(두꺼운 소재 및 푸른 띠 디테일)이 바람에 흩날리는 묘사가 프롬프트와 매우 일치합니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "로케이션과 전경 앵글은 충실히 구현되었으나, 붉은 의상의 질감이 시폰처럼 얇게 변형되었고 레퍼런스의 디테일(푸른 띠 등)이 누락되었습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 로케이션과 앵글을 잘 유지했으며, 레퍼런스의 붉은 무당복(두꺼운 소재 및 푸른 띠 디테일)이 바람에 흩날리는 묘사가 프롬프트와 매우 일치합니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "로케이션과 전경 앵글은 충실히 구현되었으나, 붉은 의상의 질감이 시폰처럼 얇게 변형되었고 레퍼런스의 디테일(푸른 띠 등)이 누락되었습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지정된 로케이션의 실제 지형을 정확히 유지하면서, 벼랑 끝에 서서 붉은 무복이 흩날리는 인물의 전경을 지시문에 맞게 훌륭하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "인물과 의상이 흩날리는 전경의 구도는 좋으나, 인물이 서 있는 절벽 끝부분의 지형이 레퍼런스 사진과 다르게 임의로 돌출된 형태로 변형되었습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 로케이션의 실제 지형을 정확히 유지하면서, 벼랑 끝에 서서 붉은 무복이 흩날리는 인물의 전경을 지시문에 맞게 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "인물과 의상이 흩날리는 전경의 구도는 좋으나, 인물이 서 있는 절벽 끝부분의 지형이 레퍼런스 사진과 다르게 임의로 돌출된 형태로 변형되었습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 12,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "지정된 로케이션과 앵글을 잘 유지했으며, 레퍼런스의 붉은 무당복(두꺼운 소재 및 푸른 띠 디테일)이 바람에 흩날리는 묘사가 프롬프트와 매우 일치합니다."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "로케이션과 전경 앵글은 충실히 구현되었으나, 붉은 의상의 질감이 시폰처럼 얇게 변형되었고 레퍼런스의 디테일(푸른 띠 등)이 누락되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해안절벽_정상_8b9a7f.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1010229>"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S95sh1__bgfirst_bg.png",
   "bg_asset_id": "c9ca11ca-5b62-4c6d-9b2f-e82406a3b54f",
   "bg_record_key": "S95sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "해안절벽_정상",
   "groupbg_asset_id": "61c6f3c2-be55-452d-853a-13950342d6d5"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S95sh10": {
  "input_fingerprint": "e08a1d99aaef1e26",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, clear sky.\n\nSHOT TEXT (authoritative, Korean): 칼날이 고산들의 등판에 깊숙이 꽂혀 있는 찰나, 강영란(흰 머리 여자)이 칼자루를 쥐고 체중을 실은 미드임팩트 모습.\n\nLOCATION (lock): On the cliff-top ground immediately behind the man facing the sea. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the high cliff, blue sea, bright sky, wind, and red ceremonial clothing. Exclude the earlier peaceful seaward stance and show the stabbing impact from behind.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul’s prior injuries persist as Youngran drives a knife deeply into his back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 8,
   "B": 3,
   "A": 2
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "카메라 위치와 남자의 등판에 칼을 꽂고 체중을 싣는 핵심 액션을 정확한 구도와 자세로 완벽하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "여성이 앞에서 찌르는데 칼날이 등 뒤로 뚫고 나오는 물리적으로 불가능하고 모순된 연출이 있어 심각한 오류입니다."
   },
   {
    "label": "A",
    "score": 2,
    "verdict_ko": "찔리는 대상(남성)과 찌르는 액션 자체가 화면에 아예 누락되어 프롬프트의 핵심 지시를 완전히 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S95sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1010229>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "칼을 쥔 영란의 오른손 뒤쪽(몸 방향)으로도 뾰족한 칼날이 튀어나와 있어 칼의 형태가 비정상적입니다.",
     "fix_en": "Remove the extra blade extending backwards from the hand towards the woman, leaving only a proper handle in her grip.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "칼자루를 쥔 오른손의 엄지와 손가락 관절 위치가 뒤틀려 해부학적으로 어색합니다.",
     "fix_en": "Redraw the right hand gripping the knife so the fingers and thumb are positioned naturally for a forward stab.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "프롬프트에 명시된 찔리는 남자(산들)의 '이전 부상(prior injuries)' 흔적이 전혀 없고 옷이 깨끗합니다.",
     "fix_en": "Add visible dirt, tear marks, and blood stains to the back of the man's red clothing to reflect prior injuries.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the extra blade extending backwards from the hand towards the woman, leaving only a proper handle in her grip.\n- Redraw the right hand gripping the knife so the fingers and thumb are positioned naturally for a forward stab.\n- Add visible dirt, tear marks, and blood stains to the back of the man's red clothing to reflect prior injuries.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S95sh1"
  }
 },
 "S95sh18": {
  "input_fingerprint": "5d611d0f7f745014",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, clear sky.\n\nSHOT TEXT (authoritative, Korean): 황망함과 공포가 뒤섞인 표정으로 절벽 아래 허공에서 옷자락이 위로 솟구치며 추락 중인 고산들의 미드액션 전신.\n\nLOCATION (lock): In open air below the high sea-cliff edge, above the rocks and water. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the cliff edge, open sea, bright daylight, and strong wind. Exclude the stabbing figures remaining on the ledge and show the wounded man falling below.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul falls with deep stab wounds through his back and chest, in addition to his earlier shoulder wound.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 4,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "샷 텍스트와 이전 샷 지시를 정확히 우선하여, 추락하는 남성의 전신과 흩날리는 옷자락을 훌륭하게 연출했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "샷 텍스트가 명시한 남성 대신 인물 제약에만 맞춰 노년 여성을 묘사해 상위 우선순위를 위반했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "바다와 하늘의 수평선이 90도로 뒤틀린 물리적으로 불가능한 배경 구조를 그려 하드 위반에 해당합니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S95sh10_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1445120>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "인물이 거꾸로 추락하고 있음에도 한복의 하얀 깃(동정)이 몸에 비해 위아래가 뒤집혀 그려져 있어, V자 넥라인이 가슴(위쪽)이 아닌 턱(아래쪽)을 향하고 있습니다.",
     "fix_en": "Redraw the white hanbok collar so it correctly wraps around the back of the neck, with the 'V' shape pointing upwards toward the character's waist.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the white hanbok collar so it correctly wraps around the back of the neck, with the 'V' shape pointing upwards toward the character's waist.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S95sh10"
  }
 },
 "S96sh1::bgfirst_bg": {
  "input_fingerprint": "142a230e20b5fa4b",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 동굴 바닥에 쇠사슬로 묶인 채 고개를 푹 숙인 김지후(강수리영)의 절망적인 전신.\n\nLOCATION (lock): On the dark cave floor at the chained prison area during daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Place the static camera at floor height directly in front of 김지후(강수리영), far enough back to contain her full seated body and the visible run of chain. She occupies the lower middle of the wide frame with her bowed head and collapsed posture compressed against the surrounding cave space.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: cave floor (김지후(강수리영) is seated on it); used as Creates the low foreground plane and emphasizes her inability to rise; chain (Fastened around and restraining her) — Its links extend from 김지후(강수리영)'s restraints across the cave floor; used as Traces the restraint outward from her body and supplies a leading line at floor level; cave interior (Dark); used as Encloses the figure and limits background depth.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 동굴 바닥에 쇠사슬로 묶인 채 고개를 푹 숙인 김지후(강수리영)의 절망적인 전신.\n\nLOCATION (lock): On the dark cave floor at the chained prison area during daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Place the static camera at floor height directly in front of 김지후(강수리영), far enough back to contain her full seated body and the visible run of chain. She occupies the lower middle of the wide frame with her bowed head and collapsed posture compressed against the surrounding cave space.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: cave floor (김지후(강수리영) is seated on it); used as Creates the low foreground plane and emphasizes her inability to rise; chain (Fastened around and restraining her) — Its links extend from 김지후(강수리영)'s restraints across the cave floor; used as Traces the restraint outward from her body and supplies a leading line at floor level; cave interior (Dark); used as Encloses the figure and limits background depth.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S96sh1__bgfirst_bg.png",
  "asset_id": "41d48f6a-b722-425b-9016-93fc6621410c",
  "input_asset_ids": [
   "b5a000e8-7899-4a48-b053-da46af51a7ce",
   "0f54c0d2-2836-40af-b48b-c7fa10a1979e"
  ]
 },
 "S96sh1": {
  "input_fingerprint": "9a30d2d54e5a6010",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어두운 동굴 바닥에 쇠사슬로 묶인 채 고개를 푹 숙인 김지후(강수리영)의 절망적인 전신.\n\nLOCATION (lock): On the dark cave floor at the chained prison area during daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is seated in a defeated slump on the cave floor with her body and limbs secured by heavy chains, her head bowed and her gaze lowered toward the ground.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains seated in a defeated slump on the cave floor with her body and limbs secured by heavy chains, her head bowed and her gaze lowered toward the ground; Inwoo remains chained to the opposite cave wall; Inwoo’s face remains battered and Jihu’s pendant remains in Youngran’s possession.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어두운 동굴 바닥에 쇠사슬로 묶인 채 고개를 푹 숙인 김지후(강수리영)의 절망적인 전신.\n\nLOCATION (lock): On the dark cave floor at the chained prison area during daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is seated in a defeated slump on the cave floor with her body and limbs secured by heavy chains, her head bowed and her gaze lowered toward the ground.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains seated in a defeated slump on the cave floor with her body and limbs secured by heavy chains, her head bowed and her gaze lowered toward the ground; Inwoo remains chained to the opposite cave wall; Inwoo’s face remains battered and Jihu’s pendant remains in Youngran’s possession.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어두운 동굴 바닥에 쇠사슬로 묶인 채 고개를 푹 숙인 김지후(강수리영)의 절망적인 전신.\n\nLOCATION (lock): On the dark cave floor at the chained prison area during daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is seated in a defeated slump on the cave floor with her body and limbs secured by heavy chains, her head bowed and her gaze lowered toward the ground.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains seated in a defeated slump on the cave floor with her body and limbs secured by heavy chains, her head bowed and her gaze lowered toward the ground; Inwoo remains chained to the opposite cave wall; Inwoo’s face remains battered and Jihu’s pendant remains in Youngran’s possession.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S96sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S96sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B02.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "샷 텍스트에 명시되지 않은 인물이 배경에 등장한 위반이 있으나, 벽의 고리와 논리적으로 연결된 쇠사슬, 소품(펜던트), 절망적인 자세 묘사가 A보다 우수합니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "샷 텍스트에 없는 인물이 배경에 포함된 위반이 있으며, 벽의 사슬은 그대로 둔 채 출처를 알 수 없는 사슬이 인물을 감싸고 있어 공간적 개연성이 떨어집니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "샷 텍스트에 명시되지 않은 인물이 배경에 등장한 위반이 있으나, 벽의 고리와 논리적으로 연결된 쇠사슬, 소품(펜던트), 절망적인 자세 묘사가 A보다 우수합니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "샷 텍스트에 없는 인물이 배경에 포함된 위반이 있으며, 벽의 사슬은 그대로 둔 채 출처를 알 수 없는 사슬이 인물을 감싸고 있어 공간적 개연성이 떨어집니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 '전신' 샷에 맞게 인물을 화면에 적절한 크기로 담아냈으며, 레퍼런스의 의상과 동굴의 구조를 훌륭하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "카메라가 지나치게 뒤로 물러나 인물이 너무 작게 렌더링되어 '전신' 샷의 프레이밍 기준(규칙 2)을 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 '전신' 샷에 맞게 인물을 화면에 적절한 크기로 담아냈으며, 레퍼런스의 의상과 동굴의 구조를 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "카메라가 지나치게 뒤로 물러나 인물이 너무 작게 렌더링되어 '전신' 샷의 프레이밍 기준(규칙 2)을 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 8,
     "B": 13
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 8,
   "B": 13
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "샷 텍스트에 명시되지 않은 인물이 배경에 등장한 위반이 있으나, 벽의 고리와 논리적으로 연결된 쇠사슬, 소품(펜던트), 절망적인 자세 묘사가 A보다 우수합니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "샷 텍스트에 없는 인물이 배경에 포함된 위반이 있으며, 벽의 사슬은 그대로 둔 채 출처를 알 수 없는 사슬이 인물을 감싸고 있어 공간적 개연성이 떨어집니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B02.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "샷 텍스트에 없는 인물(배경 오른쪽의 남자)이 추가되었습니다.",
     "fix_en": "Remove the man sitting in the background on the right, leaving only the empty cave floor and wall.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "설정상 다른 인물이 소지해야 할 펜던트를 지후가 목에 걸고 있습니다.",
     "fix_en": "Remove the bone pendant necklace from Jihu's chest, leaving only the plain texture of her dark green sweater.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "지후의 두 손이 다리 위로 떠서 마주 잡고 있어 무기력하게 축 늘어진 상태(중력 적용)가 아닙니다.",
     "fix_en": "Modify Jihu's hands so they lie completely limp, unclasped, and flat against her thighs, fully supported by gravity.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the man sitting in the background on the right, leaving only the empty cave floor and wall.\n- Remove the bone pendant necklace from Jihu's chest, leaving only the plain texture of her dark green sweater.\n- Modify Jihu's hands so they lie completely limp, unclasped, and flat against her thighs, fully supported by gravity.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S96sh1__bgfirst_bg.png",
   "bg_asset_id": "41d48f6a-b722-425b-9016-93fc6621410c",
   "bg_record_key": "S96sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S96sh4": {
  "input_fingerprint": "6c9281471bda205e",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)이 바닥에 주저앉은 김지후(강수리영)를 내려다보며 만족스러운 미소를 머금은 채 서 있는 측면.\n\nLOCATION (lock): Inside the cave prison beside the seated captive, illuminated by a carried lamp. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the dark cave floor, rough stone walls, chains, and the captive woman's seated position. Exclude the empty despair pose and add the elder standing over her with a lamp.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu and Inwoo remain chained in the cave, with Inwoo’s facial injuries unchanged.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 5,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "측면 구도와 지시된 표정을 잘 담아냈으며, 레퍼런스의 배경 요소와 포즈 설정을 가장 안정적으로 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "인물 배치와 조명은 양호하나, 노인의 손가락이 램프 손잡이를 투과하는 물리적 오류가 발생해 감점되었습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "지시된 상황은 표현되었으나, 노인의 자세가 경직되어 있고 손가락이 램프 금속부와 뭉개지는 치명적 오류가 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S96sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 샷의 신체 자세와 접촉점을 정확히 유지하라는 지시를 위반하여, 포로의 양손이 무릎 위에 모여 있지 않고 떨어져 바닥과 허벅지에 놓여 있습니다.",
     "fix_en": "Reposition the captive's arms and hands so they rest together in her lap, exactly matching the bound pose of the reference image.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "포로의 허벅지 위에 놓인 왼쪽 손의 손가락이 기형적이며 바지 직물과 융합되어 해부학적으로 뭉개져 있습니다.",
     "fix_en": "Redraw the captive's left hand to feature clean, anatomically correct fingers.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "랜턴을 쥐고 있는 노인의 오른손 손가락이 램프의 금속 손잡이와 녹아내리듯 융합되어 해부학적 오류가 있습니다.",
     "fix_en": "Redraw the fingers of the elder's right hand so they distinctly and naturally grip the lantern handle without melting into the metal.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Reposition the captive's arms and hands so they rest together in her lap, exactly matching the bound pose of the reference image.\n- Redraw the captive's left hand to feature clean, anatomically correct fingers.\n- Redraw the fingers of the elder's right hand so they distinctly and naturally grip the lantern handle without melting into the metal.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S96sh1"
  }
 },
 "S97sh1": {
  "input_fingerprint": "d5426b252bd8bfb7",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 하얀 사슴뿔 창끝이 그분(검은 연기 형체)의 등판을 정면으로 뚫고 튀어나온 찰나의 클로즈업.\n\nLOCATION (lock): On the bloodied ground directly before the old wooden storehouse in a winter reed field. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The black figure remains transfixed through the heart by the long white-antler-tipped spear.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 3,
   "B": 2,
   "C": 8
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "창이 인물의 등을 뚫지 않고 허공에 떠 있는 물리적 오류(Hard Violation)가 발생했으며, 클로즈업 프레이밍도 아닙니다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "인물을 구체적인 인간이 아닌 텍스트 그대로의 검은 연기로 렌더링하여 캐릭터 묘사 지침을 심각하게 위반했습니다."
   },
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "사슴뿔 창끝이 등을 뚫고 나온 클로즈업 샷을 정확히 구현했으며, 비유적인 인물 묘사를 검은 옷을 입은 실제 인간으로 올바르게 해석했습니다."
   }
  ],
  "refs": [],
  "critique": {
   "issues": [
    {
     "issue_ko": "사슴뿔이 인체의 등판 중앙을 뚫고 나온 형태가 아니라, 잘린 원통형 천(목이나 소매 단면)에서 빠져나온 것처럼 보입니다.",
     "fix_en": "Modify the black fabric mass around the antler's base to clearly resemble the flat surface of a human back, ensuring the antler pierces directly through the center of the garment.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Modify the black fabric mass around the antler's base to clearly resemble the flat surface of a human back, ensuring the antler pierces directly through the center of the garment.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S97sh5": {
  "input_fingerprint": "5843b5b36aadeb4f",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 목을 부여잡은 조선인 촌주의 두 눈동자가 섬뜩한 황금빛으로 번쩍이는 찰나의 클로즈업.\n\nLOCATION (lock): Immediately outside the reed-field storehouse beside the fading smoke figure. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the daylight reed-field warehouse exterior, fallen soldiers, and fading smoke figure. Exclude the spear-tip close-up and focus on the village leader's changing eyes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The black figure has vanished after dying on the spear, while the village chief’s eyes retain the transferred golden glow.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 6,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "클로즈업 샷의 구도를 잘 살렸으며, 목을 부여잡은 포즈와 황금빛 눈동자, 그리고 흩어지는 연기 형상을 가장 정확하게 묘사했습니다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "쓰러진 병사들은 묘사되었으나, 프롬프트에 명시된 '사라져가는 연기 형상'이 화면에서 완전히 누락되었습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "레퍼런스 이미지에서 이미 사라진 것으로 설정된 검은 존재의 특이한 질감의 옷을 촌장에게 잘못 입혀 설정 오류를 범했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S97sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "촌주의 홍채가 물리적으로 노랗게 변형되어 발광하고 있어, 모든 살아있는 인물의 눈은 해부학적으로 정상이어야 하며 비유적 표현이나 광채는 조명으로 연출해야 한다는 규칙을 위반했습니다.",
     "fix_en": "Restore the glowing yellow irises to anatomically normal dark brown human eyes, and represent the 'golden glow' by adding a warm golden light reflection onto the surface of the natural eyes.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore the glowing yellow irises to anatomically normal dark brown human eyes, and represent the 'golden glow' by adding a warm golden light reflection onto the surface of the natural eyes.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S97sh1"
  }
 },
 "S97sh6": {
  "input_fingerprint": "eee0cd4fce9973db",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 낡은 창고 앞 흙바닥에서 과거의 조선인 육혈귀 생존자들이 피투성이가 된 일본군 병사들의 시신에 짐승처럼 이빨을 깊게 박아넣은 찰나의 참혹한 광경.\n\nLOCATION (lock): On the dirt before the reed-field storehouse, among the bloodied soldiers' bodies. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the rough warehouse, reed field, daylight, dead soldiers, and blood-soaked ground. Exclude the dying smoke figure and show the survivors feeding on the bodies.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): The dead Japanese soldiers are sprawled across the dirt outside the wooden storehouse with their torsos and slack limbs collapsed against the ground while the survivors crouch over and bite into their bodies.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The dead Japanese soldiers are sprawled across the dirt outside the wooden storehouse with their torsos and slack limbs collapsed against the ground while the survivors crouch over and bite into their bodies.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 3,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "생존자가 시신을 물어뜯는 핵심 행동과 피투성이 묘사를 지문과 레퍼런스에 맞게 가장 정확히 구현했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "시신을 물어뜯는 찰나의 행동이 누락되었으며, 요구된 피투성이 상태가 전혀 표현되지 않았습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "행동이 뒷모습에 가려져 불분명하며, 생존자들의 복장과 구도가 다소 단조롭고 반복적으로 보입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S97sh5_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "전경의 생존자가 시신의 살점에 이빨을 깊게 박아넣지 않고 입만 벌린 채 허공에 떠 있습니다.",
     "fix_en": "Lower the foreground survivor's face so his mouth and teeth are sunk deeply into the dead soldier's neck.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "전경에 누워있는 일본군 시신의 목이 해부학적으로 비정상적으로 길게 늘어나 기괴한 형태를 띠고 있습니다.",
     "fix_en": "Shorten the foreground dead soldier's bloody neck to normal human proportions, bringing the head closer to the torso.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "전경의 생존자가 레퍼런스 이미지(얇은 망건)와 달리 두꺼운 검은색 천을 머리에 두르고 있습니다.",
     "fix_en": "Replace the thick black cloth headband on the foreground survivor with a thin horsehair headband (manggeon) that matches the reference image.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Lower the foreground survivor's face so his mouth and teeth are sunk deeply into the dead soldier's neck.\n- Shorten the foreground dead soldier's bloody neck to normal human proportions, bringing the head closer to the torso.\n- Replace the thick black cloth headband on the foreground survivor with a thin horsehair headband (manggeon) that matches the reference image.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S97sh5"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S98sh2::bgfirst_bg": {
  "input_fingerprint": "3cbb88c5fd8f0c47",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 눈물이 고인 눈으로 강영란(흰 머리 여자)을 매섭게 노려보며 분노에 차 입을 크게 벌려 소리치는 김지후(강수리영)의 얼굴.\n\nLOCATION (lock): Inside the cave prison, facing the lamp-lit visitor from the chained holding area.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track laterally into an eye-level frontal close position on 김지후(강수리영), stopping with her tear-filled face occupying most of the frame while retaining a narrow trace of chains and cave context below. Her head thrusts toward 강영란(흰 머리 여자)'s off-screen position, mouth open mid-shout and eyes locked just past the lens rather than into it.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: chain (Still restraining her) — A partial run of links leads downward from 김지후(강수리영)'s body; used as Provides a small lower-frame reminder that the shouting figure remains restrained; cave interior (Visible only as limited background context); used as Falls out of focus behind the close face.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 눈물이 고인 눈으로 강영란(흰 머리 여자)을 매섭게 노려보며 분노에 차 입을 크게 벌려 소리치는 김지후(강수리영)의 얼굴.\n\nLOCATION (lock): Inside the cave prison, facing the lamp-lit visitor from the chained holding area.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track laterally into an eye-level frontal close position on 김지후(강수리영), stopping with her tear-filled face occupying most of the frame while retaining a narrow trace of chains and cave context below. Her head thrusts toward 강영란(흰 머리 여자)'s off-screen position, mouth open mid-shout and eyes locked just past the lens rather than into it.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: chain (Still restraining her) — A partial run of links leads downward from 김지후(강수리영)'s body; used as Provides a small lower-frame reminder that the shouting figure remains restrained; cave interior (Visible only as limited background context); used as Falls out of focus behind the close face.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S98sh2__bgfirst_bg.png",
  "asset_id": "b3e8285d-e226-4eea-8244-f3181a6531f1",
  "input_asset_ids": [
   "318eb9e6-dc40-4566-a147-1a01a59549fb",
   "0f54c0d2-2836-40af-b48b-c7fa10a1979e"
  ]
 },
 "S98sh2": {
  "input_fingerprint": "53a40b610053fdb6",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 눈물이 고인 눈으로 강영란(흰 머리 여자)을 매섭게 노려보며 분노에 차 입을 크게 벌려 소리치는 김지후(강수리영)의 얼굴.\n\nLOCATION (lock): Inside the cave prison, facing the lamp-lit visitor from the chained holding area. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She remains seated on the cave floor with her limbs secured by heavy chains, but raises her head and leans forward to glare and shout at Youngran.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu and Inwoo remain chained in the cave; Jihu’s pendant remains under Youngran’s control.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 눈물이 고인 눈으로 강영란(흰 머리 여자)을 매섭게 노려보며 분노에 차 입을 크게 벌려 소리치는 김지후(강수리영)의 얼굴.\n\nLOCATION (lock): Inside the cave prison, facing the lamp-lit visitor from the chained holding area. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She remains seated on the cave floor with her limbs secured by heavy chains, but raises her head and leans forward to glare and shout at Youngran.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu and Inwoo remain chained in the cave; Jihu’s pendant remains under Youngran’s control.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 눈물이 고인 눈으로 강영란(흰 머리 여자)을 매섭게 노려보며 분노에 차 입을 크게 벌려 소리치는 김지후(강수리영)의 얼굴.\n\nLOCATION (lock): Inside the cave prison, facing the lamp-lit visitor from the chained holding area. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She remains seated on the cave floor with her limbs secured by heavy chains, but raises her head and leans forward to glare and shout at Youngran.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu and Inwoo remain chained in the cave; Jihu’s pendant remains under Youngran’s control.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S98sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S98sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B02.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:736787>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지문이 요구한 '얼굴'이라는 숏 크기에 맞춰 타 인물이나 신체를 배제하고 얼굴에 집중한 프레이밍을 정확히 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "'얼굴'에 집중하라는 지문을 벗어나 상반신, 팔, 다른 인물까지 보이도록 프레임을 임의로 넓혀 숏 크기 규칙을 위반했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지문이 요구한 '얼굴'이라는 숏 크기에 맞춰 타 인물이나 신체를 배제하고 얼굴에 집중한 프레이밍을 정확히 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "'얼굴'에 집중하라는 지문을 벗어나 상반신, 팔, 다른 인물까지 보이도록 프레임을 임의로 넓혀 숏 크기 규칙을 위반했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지시문에 명시된 '흰 머리 여자(강영란)'를 오버 더 숄더로 배치하여 노려보는 대상을 명확히 했으며, 지후의 분노한 표정과 눈물 연출이 우수합니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "지후의 표정과 동굴 배경은 잘 표현되었으나, 텍스트가 요구한 '흰 머리 여자'가 화면에 전혀 등장하지 않아 상호작용이 누락되었습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지시문에 명시된 '흰 머리 여자(강영란)'를 오버 더 숄더로 배치하여 노려보는 대상을 명확히 했으며, 지후의 분노한 표정과 눈물 연출이 우수합니다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "지후의 표정과 동굴 배경은 잘 표현되었으나, 텍스트가 요구한 '흰 머리 여자'가 화면에 전혀 등장하지 않아 상호작용이 누락되었습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 12
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 12,
   "B": 12
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지문이 요구한 '얼굴'이라는 숏 크기에 맞춰 타 인물이나 신체를 배제하고 얼굴에 집중한 프레이밍을 정확히 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "'얼굴'에 집중하라는 지문을 벗어나 상반신, 팔, 다른 인물까지 보이도록 프레임을 임의로 넓혀 숏 크기 규칙을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B02.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "캐릭터의 시선이 스케치의 화살표 방향(화면 오른쪽)을 향하지 않고 카메라 정면을 바라보고 있습니다.",
     "fix_en": "Shift her pupils to look sharply toward the right side of the frame, exactly matching the gaze arrow from the layout sketch.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Shift her pupils to look sharply toward the right side of the frame, exactly matching the gaze arrow from the layout sketch.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S98sh2__bgfirst_bg.png",
   "bg_asset_id": "b3e8285d-e226-4eea-8244-f3181a6531f1",
   "bg_record_key": "S98sh2::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S98sh5": {
  "input_fingerprint": "f63085aabd5f0e70",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 쇠사슬에 묶인 채 어이없다는 듯 미간을 잔뜩 찌푸리며 강영란(흰 머리 여자)을 향해 목에 핏줄이 불거진 채 입을 벌려 소리치는 정인우의 전신.\n\nLOCATION (lock): At the opposite wall of the cave prison, where the captive remains chained. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the cave chamber, wall chains, bound pair, and dim light. Exclude the younger woman's close-up emphasis and show the bound man shouting at the elder.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): He remains chained against the opposite cave wall with his torso held upright and his limbs secured, raising his head and straining his neck toward Youngran as he shouts.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains chained to the cave walls, and Inwoo remains chained against the opposite cave wall with his torso held upright and his limbs secured, raising his head and straining his neck toward Youngran as he shouts, and Inwoo’s face remains battered.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지정된 '전신' 프레이밍과 참조 이미지의 복장을 정확히 재현했으며, 지시된 모든 상황 요소를 충실히 반영했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "요구된 '전신' 프레이밍을 위반하여 상반신만 렌더링되었고, 캐릭터의 지정된 아우터 복장도 누락되었습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "요구된 '전신' 프레이밍을 위반하여 신체 하단이 잘렸으며, 캐릭터의 지정된 복장(재킷, 조끼)이 완전히 누락되었습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S98sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903809>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지문에 언급되지 않은 인물(배경에 앉아 있는 젊은 여성)이 프레임에 포함되어 있습니다.",
     "fix_en": "Remove the young woman sitting in the background.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "정인우의 상체가 꼿꼿하게 유지(held upright)되지 않고 앞으로 크게 기울어진 런지 자세를 취하고 있습니다.",
     "fix_en": "Adjust the man's pose so his torso is held upright against the wall instead of leaning forward.",
     "needs_regeneration": false,
     "unfixable": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the young woman sitting in the background.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S98sh2"
  }
 },
 "S98sh6": {
  "input_fingerprint": "dc64c488ffa38f97",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 정인우의 비난에 시니컬한 미소를 지은 채 김지후(강수리영)를 차갑게 내려다보는 강영란(흰 머리 여자)의 얼굴.\n\nLOCATION (lock): Beside the seated captive within the dim, lamp-lit cave chamber. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the rough cave interior, restrained captives, lamp light, and the elder's standing position. Exclude the man's shouting emphasis and focus on the elder's cynical expression.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu and Inwoo remain chained; Jihu’s antler-bone pendant remains in Youngran’s possession.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 5,
   "C": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "샷 텍스트가 지시한 '얼굴' 중심의 밀착된 프레이밍과 내려다보는 시선, 시니컬한 표정을 가장 정확하게 포착했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "이전 샷의 '서 있는 자세'를 유지하라는 지시를 어기고 쪼그려 앉아 있으며, 얼굴 중심의 프레이밍 지시에서 벗어났습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "고정되어야 할 배경 인물의 포즈가 이전 샷과 다르게 변형되었으며, 얼굴 집중 프레이밍을 어기고 너무 넓게 씬을 잡았습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S98sh5_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1010229>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트의 인물 허용 규칙에 따라 '강영란(흰 머리 여자)' 한 명만 화면에 나타나야 하지만, 전경 우측에 다른 남자의 뒷모습이 포함되어 있습니다.",
     "fix_en": "Remove the man in the right foreground, replacing him with the cave background and the woman's black clothing so she is the only person in frame.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "캐릭터 설정에 '가지런히 틀어 묶은 머리'로 명시되어 있으나, 여자의 흰 머리가 길게 풀려 있는 상태입니다.",
     "fix_en": "Change the woman's loose, flowing white hair to be neatly tied up.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the man in the right foreground, replacing him with the cave background and the woman's black clothing so she is the only person in frame.\n- Change the woman's loose, flowing white hair to be neatly tied up.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S98sh5"
  }
 },
 "S99sh1::bgfirst_bg": {
  "input_fingerprint": "c0819c730671e1fc",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 밝은 지구대 안에서 구창모가 의자에 느긋하게 기대앉아 핸드폰 화면을 들여다보는 모습.\n\nLOCATION (lock): Inside a brightly lit mainland police substation, at a desk and chair.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: A static wide view from seated eye level places 구창모 reclining with his chair and immediate workspace clearly legible around him. He occupies one side of the midground, his raised phone kept at a natural scale while open room space anticipates the coming pan toward the nearby officer.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 의자 (occupied) — Its back faces diagonally toward the camera while supporting 구창모; used as Supports 구창모's relaxed posture and establishes his position within the workspace; 업무 책상 (in use) — The near edge and top plane are visible from the seated-height camera position; used as Anchors the immediate police-station workspace beside the seated figure; 핸드폰 화면 (on, displaying a game) — The active game screen faces 구창모 and is seen obliquely by the camera; used as Held within 구창모's hands as the initial point of attention without overwhelming the wider composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 밝은 지구대 안에서 구창모가 의자에 느긋하게 기대앉아 핸드폰 화면을 들여다보는 모습.\n\nLOCATION (lock): Inside a brightly lit mainland police substation, at a desk and chair.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: A static wide view from seated eye level places 구창모 reclining with his chair and immediate workspace clearly legible around him. He occupies one side of the midground, his raised phone kept at a natural scale while open room space anticipates the coming pan toward the nearby officer.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 의자 (occupied) — Its back faces diagonally toward the camera while supporting 구창모; used as Supports 구창모's relaxed posture and establishes his position within the workspace; 업무 책상 (in use) — The near edge and top plane are visible from the seated-height camera position; used as Anchors the immediate police-station workspace beside the seated figure; 핸드폰 화면 (on, displaying a game) — The active game screen faces 구창모 and is seen obliquely by the camera; used as Held within 구창모's hands as the initial point of attention without overwhelming the wider composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S99sh1__bgfirst_bg.png",
  "asset_id": "a43fc8d5-82d9-4da8-b732-aa8242569245",
  "input_asset_ids": [
   "dff9730e-f2a2-454d-a541-e1275378a6de",
   "e784780d-4838-4f83-8abe-8779c9c35062"
  ]
 },
 "S99sh1": {
  "input_fingerprint": "5e6c64203e09aef2",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 지구대 안에서 구창모가 의자에 느긋하게 기대앉아 핸드폰 화면을 들여다보는 모습.\n\nLOCATION (lock): Inside a brightly lit mainland police substation, at a desk and chair. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 지구대 안에서 구창모가 의자에 느긋하게 기대앉아 핸드폰 화면을 들여다보는 모습.\n\nLOCATION (lock): Inside a brightly lit mainland police substation, at a desk and chair. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 지구대 안에서 구창모가 의자에 느긋하게 기대앉아 핸드폰 화면을 들여다보는 모습.\n\nLOCATION (lock): Inside a brightly lit mainland police substation, at a desk and chair. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S99sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S99sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L03B02.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "핸드폰의 화면이 카메라를 향하고 있어 인물이 기기의 뒷면을 보는 물리적 오류가 있으며, 레퍼런스의 카메라 구도를 그대로 복사하여 지침을 크게 위반했습니다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "레퍼런스의 구도를 복사하지 않고 지정된 장소 안에서 새로운 앵글을 잘 구현했으며, 핸드폰의 기능적 면(화면)이 인물의 눈을 향하도록 올바르게 연출했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "핸드폰의 화면이 카메라를 향하고 있어 인물이 기기의 뒷면을 보는 물리적 오류가 있으며, 레퍼런스의 카메라 구도를 그대로 복사하여 지침을 크게 위반했습니다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "레퍼런스의 구도를 복사하지 않고 지정된 장소 안에서 새로운 앵글을 잘 구현했으며, 핸드폰의 기능적 면(화면)이 인물의 눈을 향하도록 올바르게 연출했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "레퍼런스의 카메라 구도를 복사하지 말라는 지침을 준수하여, 지정된 공간 내에서 새로운 앵글로 인물의 동작을 자연스럽게 연출했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "레퍼런스 사진의 카메라 구도를 절대 복사하지 말라는 명시적인 지침을 어기고 원본과 동일한 프레이밍을 생성했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "레퍼런스의 카메라 구도를 복사하지 말라는 지침을 준수하여, 지정된 공간 내에서 새로운 앵글로 인물의 동작을 자연스럽게 연출했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "레퍼런스 사진의 카메라 구도를 절대 복사하지 말라는 명시적인 지침을 어기고 원본과 동일한 프레이밍을 생성했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 7,
     "B": 15
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 7,
   "B": 15
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "핸드폰의 화면이 카메라를 향하고 있어 인물이 기기의 뒷면을 보는 물리적 오류가 있으며, 레퍼런스의 카메라 구도를 그대로 복사하여 지침을 크게 위반했습니다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "레퍼런스의 구도를 복사하지 않고 지정된 장소 안에서 새로운 앵글을 잘 구현했으며, 핸드폰의 기능적 면(화면)이 인물의 눈을 향하도록 올바르게 연출했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L03B02.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "모니터 왼쪽 아래에 붙은 메모, 뒷벽의 파란색 표지판 및 벽에 걸린 종이, 오른쪽 책상 위 서류 등에 가짜 텍스트가 쓰여 있습니다.",
     "fix_en": "Remove all text and pseudo-text from the white note on the left monitor, the blue sign and papers on the back wall, and the documents on the right desk, leaving those surfaces blank.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all text and pseudo-text from the white note on the left monitor, the blue sign and papers on the back wall, and the documents on the right desk, leaving those surfaces blank.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S99sh1__bgfirst_bg.png",
   "bg_asset_id": "a43fc8d5-82d9-4da8-b732-aa8242569245",
   "bg_record_key": "S99sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S99sh4": {
  "input_fingerprint": "5810519c46a0dbd0",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 수화기를 귀에 댄 구창모가 놀란 눈으로 허공을 응시하며 완전히 얼어붙은 얼굴 클로즈업.\n\nLOCATION (lock): At the desk telephone inside the daylight police substation. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the bright small police office, desk, chair, phone, and daytime fluorescent light. Exclude the earlier mobile-game pose and show the officer frozen with the receiver at his ear.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "요구된 얼굴 클로즈업과 놀란 표정을 훌륭히 연출했으며, 레퍼런스의 배경 요소들을 텍스트 위반 없이 안정적으로 유지했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "캐릭터의 상태와 배경은 지시를 잘 따랐으나, 프레이밍이 클로즈업이 아닌 미디엄 샷에 가깝게 넓어 몰입도가 다소 떨어집니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "클로즈업 프레이밍은 가장 정확하지만, 배경에 레퍼런스에 존재하지 않는 임의의 영문 텍스트(경찰 간판)가 생성되어 심각한 규칙 위반입니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S99sh1_sel.png"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S99sh1"
  }
 },
 "S100sh3::bgfirst_bg": {
  "input_fingerprint": "73425d8c2552e212",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 구창모가 거들먹거리는 표정으로 턱을 치켜들고 한 손을 허리에 얹은 채, 짙은 화장을 한 여자들 앞에 꼿꼿이 서 있는 모습.\n\nLOCATION (lock): On a mainland pier beside a parked van and the women who have just disembarked.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height from directly before the group, a static medium-wide composition places 구창모 in the foreground with his chin lifted and one hand planted at his waist, while the women form an uneven cluster beyond and beside him. The parked black van remains readable at the frame edge as evidence of their arrival, without competing with the human grouping.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 까만색 그랜드카니발 (parked) — A side and partial front or rear plane are visible from the frontal group axis; used as Provides the arrival context along one side of the ensemble frame; 해안마을 선착장 (daytime); used as Defines the outdoor working space around the assembled group.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 구창모가 거들먹거리는 표정으로 턱을 치켜들고 한 손을 허리에 얹은 채, 짙은 화장을 한 여자들 앞에 꼿꼿이 서 있는 모습.\n\nLOCATION (lock): On a mainland pier beside a parked van and the women who have just disembarked.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height from directly before the group, a static medium-wide composition places 구창모 in the foreground with his chin lifted and one hand planted at his waist, while the women form an uneven cluster beyond and beside him. The parked black van remains readable at the frame edge as evidence of their arrival, without competing with the human grouping.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 까만색 그랜드카니발 (parked) — A side and partial front or rear plane are visible from the frontal group axis; used as Provides the arrival context along one side of the ensemble frame; 해안마을 선착장 (daytime); used as Defines the outdoor working space around the assembled group.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S100sh3__bgfirst_bg.png",
  "asset_id": "03fac3f2-a20a-4dc0-8f12-f9c1a111af26",
  "input_asset_ids": [
   "fc238a76-36a9-4e85-8fe5-0733a929de2f",
   "fa927157-6e12-4428-b37d-a2a4faea6464"
  ]
 },
 "S100sh3": {
  "input_fingerprint": "2b52e3a7437b3776",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 구창모가 거들먹거리는 표정으로 턱을 치켜들고 한 손을 허리에 얹은 채, 짙은 화장을 한 여자들 앞에 꼿꼿이 서 있는 모습.\n\nLOCATION (lock): On a mainland pier beside a parked van and the women who have just disembarked. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 구창모가 거들먹거리는 표정으로 턱을 치켜들고 한 손을 허리에 얹은 채, 짙은 화장을 한 여자들 앞에 꼿꼿이 서 있는 모습.\n\nLOCATION (lock): On a mainland pier beside a parked van and the women who have just disembarked. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 구창모가 거들먹거리는 표정으로 턱을 치켜들고 한 손을 허리에 얹은 채, 짙은 화장을 한 여자들 앞에 꼿꼿이 서 있는 모습.\n\nLOCATION (lock): On a mainland pier beside a parked van and the women who have just disembarked. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S100sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S100sh3.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B01.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "거들먹거리는 표정과 턱을 치켜든 자세, 그리고 배경에 있는 여자들의 짙은 화장을 잘 구현했으나, '한 손'이라는 지시와 달리 두 손을 허리에 얹은 점이 아쉽습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "한 손을 허리에 얹은 자세는 일치하지만, 여자들의 짙은 화장 지시가 누락되었고 남성의 표정에서 거들먹거리는 느낌이 부족합니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "거들먹거리는 표정과 턱을 치켜든 자세, 그리고 배경에 있는 여자들의 짙은 화장을 잘 구현했으나, '한 손'이라는 지시와 달리 두 손을 허리에 얹은 점이 아쉽습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "한 손을 허리에 얹은 자세는 일치하지만, 여자들의 짙은 화장 지시가 누락되었고 남성의 표정에서 거들먹거리는 느낌이 부족합니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "한 손 대신 두 손을 허리춤에 얹었으나, 거들먹거리는 표정과 턱을 치켜든 자세, 뒤에 선 여자들의 짙은 화장 등 주요 묘사를 훌륭하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "거들먹거리는 표정이라기보다 단순히 허공을 응시하고 있으며, 여자들의 화장이 짙다는 설정이 시각적으로 잘 드러나지 않습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "한 손 대신 두 손을 허리춤에 얹었으나, 거들먹거리는 표정과 턱을 치켜든 자세, 뒤에 선 여자들의 짙은 화장 등 주요 묘사를 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "거들먹거리는 표정이라기보다 단순히 허공을 응시하고 있으며, 여자들의 화장이 짙다는 설정이 시각적으로 잘 드러나지 않습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 8,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 8,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "거들먹거리는 표정과 턱을 치켜든 자세, 그리고 배경에 있는 여자들의 짙은 화장을 잘 구현했으나, '한 손'이라는 지시와 달리 두 손을 허리에 얹은 점이 아쉽습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "한 손을 허리에 얹은 자세는 일치하지만, 여자들의 짙은 화장 지시가 누락되었고 남성의 표정에서 거들먹거리는 느낌이 부족합니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B01.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "구창모가 두 손을 허리에 얹고 있습니다. 프롬프트는 '한 손'만 허리에 얹을 것을 요구했습니다.",
     "fix_en": "Drop one of the man's hands from his waist so that it hangs naturally by his side, leaving only one hand resting on his hip.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Drop one of the man's hands from his waist so that it hangs naturally by his side, leaving only one hand resting on his hip.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S100sh3__bgfirst_bg.png",
   "bg_asset_id": "03fac3f2-a20a-4dc0-8f12-f9c1a111af26",
   "bg_record_key": "S100sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S100sh5": {
  "input_fingerprint": "258657b50fc26bf5",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 선착장에 정박된 낚시 어선 위에서 황봉업이 한쪽 팔을 머리 위로 번쩍 들어 올려 허공에 멈춘 채 서 있는 전신.\n\nLOCATION (lock): Aboard the fishing boat moored at the mainland pier in bright daylight. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the sunny harbor, parked van, gathered women, fishing boat, and pier. Exclude the officer's boastful pose and show the boatman waving from the vessel.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 6,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "이전 장면의 의상과 배경 요소를 완벽히 유지하면서 어선 위에서 한 팔을 번쩍 든 전신 구도를 정확하게 연출했습니다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "구도와 포즈는 적절하나 선박 표면에 텍스트가 생성되어 문자 제외 지시를 위반했으며, 배경 인물들의 형태가 미세하게 변형되었습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "주요 인물의 의상이 가죽 재킷에서 청자켓으로 완전히 바뀌어 이전 숏의 상태 유지 조건을 심각하게 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S100sh3_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지시문은 어선 위에서 한쪽 팔을 든 뱃사공(황봉업)을 요구했으나, 이전 샷에서 부두에 있던 가죽 재킷의 인물(경찰관)이 그 역할을 대신하고 있으며 배경의 원래 낚시 어선에는 아무도 없습니다.",
     "fix_en": "Remove the leather-jacketed man in the foreground, and add the boatman standing on the white and blue fishing boat in the background with one arm raised high.",
     "unfixable": true
    },
    {
     "issue_ko": "가죽 재킷의 인물이 서 있는 전경의 파란색 배와 그 아래의 물은 이전 샷의 콘크리트 부두 바닥을 훼손하고 잘못 생성된 구조물입니다.",
     "fix_en": "Remove the blue boat and the water in the foreground, restoring the solid concrete ground of the pier as established in the reference.",
     "unfixable": true
    }
   ]
  },
  "fix_skipped": true,
  "fix_skip_reason": "all_issues_unfixable",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S100sh3"
  }
 },
 "S101sh1::bgfirst_bg": {
  "input_fingerprint": "7e0505db6df78573",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 밝은 대낮의 낚시 어선 갑판 위로 화려한 옷차림의 술집 여자들과 후반부 낚싯꾼들이 발 디딜 틈 없이 빼곡하게 모여 서 있는 전경.\n\nLOCATION (lock): On the crowded open deck of a moored fishing boat in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: A static eye-level wide shot from the dock looks across the full side of the fishing boat, keeping the crowded deck layout legible from end to end. The women and fishermen overlap in dense rows across foreground and midground, with small variations in posture and head direction preventing the crowd from reading as a formal lineup.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 낚시 어선 갑판 (full of passengers) — The dock-facing side and the occupied top deck are visible across the frame; used as Contains the dense passenger arrangement and displays the full-deck crowding; 선착장 (daytime); used as Provides the stable camera platform and a narrow lower-frame boundary.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 밝은 대낮의 낚시 어선 갑판 위로 화려한 옷차림의 술집 여자들과 후반부 낚싯꾼들이 발 디딜 틈 없이 빼곡하게 모여 서 있는 전경.\n\nLOCATION (lock): On the crowded open deck of a moored fishing boat in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: A static eye-level wide shot from the dock looks across the full side of the fishing boat, keeping the crowded deck layout legible from end to end. The women and fishermen overlap in dense rows across foreground and midground, with small variations in posture and head direction preventing the crowd from reading as a formal lineup.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 낚시 어선 갑판 (full of passengers) — The dock-facing side and the occupied top deck are visible across the frame; used as Contains the dense passenger arrangement and displays the full-deck crowding; 선착장 (daytime); used as Provides the stable camera platform and a narrow lower-frame boundary.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S101sh1__bgfirst_bg.png",
  "asset_id": "b1b2ded1-8da3-4f9b-b497-849fd70db957",
  "input_asset_ids": [
   "32fc71ed-0d3e-4f53-b074-45f50e6f318c",
   "dbae1dae-10b0-4166-b7a5-b73b6a7c40b6"
  ]
 },
 "S101sh1": {
  "input_fingerprint": "498fdcf03bfb92f4",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 대낮의 낚시 어선 갑판 위로 화려한 옷차림의 술집 여자들과 후반부 낚싯꾼들이 발 디딜 틈 없이 빼곡하게 모여 서 있는 전경.\n\nLOCATION (lock): On the crowded open deck of a moored fishing boat in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (한국인 40대 중후반 남성, 호리한 얼굴, 짧은 검은 머리, 가무잡잡한 피부) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 대낮의 낚시 어선 갑판 위로 화려한 옷차림의 술집 여자들과 후반부 낚싯꾼들이 발 디딜 틈 없이 빼곡하게 모여 서 있는 전경.\n\nLOCATION (lock): On the crowded open deck of a moored fishing boat in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (한국인 40대 중후반 남성, 호리한 얼굴, 짧은 검은 머리, 가무잡잡한 피부) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 대낮의 낚시 어선 갑판 위로 화려한 옷차림의 술집 여자들과 후반부 낚싯꾼들이 발 디딜 틈 없이 빼곡하게 모여 서 있는 전경.\n\nLOCATION (lock): On the crowded open deck of a moored fishing boat in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (한국인 40대 중후반 남성, 호리한 얼굴, 짧은 검은 머리, 가무잡잡한 피부) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S101sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S101sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1284406>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L55B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1284406>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "카메라가 지정된 갑판 위에 위치하여 장소 조건을 완벽히 충족하며, 발 디딜 틈 없이 빼곡한 군중과 지정된 인물(구창모)의 외모를 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "카메라가 갑판 위가 아닌 부두에 위치해 장소 구도를 위반했으며, 인물들이 발 디딜 틈 없이 빼곡하게 모여 있다는 묘사도 부족합니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "카메라가 지정된 갑판 위에 위치하여 장소 조건을 완벽히 충족하며, 발 디딜 틈 없이 빼곡한 군중과 지정된 인물(구창모)의 외모를 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "카메라가 갑판 위가 아닌 부두에 위치해 장소 구도를 위반했으며, 인물들이 발 디딜 틈 없이 빼곡하게 모여 있다는 묘사도 부족합니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "어선 갑판 위라는 지정된 장소에서 발 디딜 틈 없이 빽빽하게 모인 군중의 모습을 텍스트에 맞게 정확히 연출했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "카메라가 갑판이 아닌 선착장에 위치하여 지정된 시점과 장소 조건을 위반했으며, 군중의 밀집도도 부족합니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "어선 갑판 위라는 지정된 장소에서 발 디딜 틈 없이 빽빽하게 모인 군중의 모습을 텍스트에 맞게 정확히 연출했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "카메라가 갑판이 아닌 선착장에 위치하여 지정된 시점과 장소 조건을 위반했으며, 군중의 밀집도도 부족합니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "카메라가 지정된 갑판 위에 위치하여 장소 조건을 완벽히 충족하며, 발 디딜 틈 없이 빼곡한 군중과 지정된 인물(구창모)의 외모를 훌륭하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "카메라가 갑판 위가 아닌 부두에 위치해 장소 구도를 위반했으며, 인물들이 발 디딜 틈 없이 빼곡하게 모여 있다는 묘사도 부족합니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L55B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1284406>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "화면 왼쪽 위 선실 외벽에 의미를 알 수 없는 깨진 파란색 글씨가 있음.",
     "fix_en": "Remove the garbled blue text from the upper left cabin wall.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "청록색 스팽글 상의를 입은 여성의 왼손 손가락이 뭉개져 있음.",
     "fix_en": "Restore the left hand of the woman in the teal sequined top to normal anatomy.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "노란색 상의를 입은 여성의 오른발과 구두가 심하게 변형되어 있음.",
     "fix_en": "Redraw the right foot and shoe of the woman in the yellow top to have correct anatomy.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "중앙 남성의 오른손 손가락 형태가 부자연스럽게 일그러져 있음.",
     "fix_en": "Fix the right hand of the man in the center to show correct finger proportions.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "얼룩말 무늬 치마를 입은 여성의 오른손 손가락이 비정상적으로 길고 기형적임.",
     "fix_en": "Fix the right hand of the woman in the zebra skirt to normal finger length.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "우측 난간 쪽에 서 있는 낚시꾼의 얼굴이 심하게 뭉개져 있음.",
     "fix_en": "Correct the face of the fisherman standing near the right railing to have normal human features.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the garbled blue text from the upper left cabin wall.\n- Restore the left hand of the woman in the teal sequined top to normal anatomy.\n- Redraw the right foot and shoe of the woman in the yellow top to have correct anatomy.\n- Fix the right hand of the man in the center to show correct finger proportions.\n- Fix the right hand of the woman in the zebra skirt to normal finger length.\n- Correct the face of the fisherman standing near the right railing to have normal human features.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S101sh1__bgfirst_bg.png",
   "bg_asset_id": "b1b2ded1-8da3-4f9b-b497-849fd70db957",
   "bg_record_key": "S101sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S101sh6": {
  "input_fingerprint": "97e246d02e1a90f7",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 황봉업(황 씨)이 선착장의 구창모 곁에 서서 넉살 좋은 미소를 띠며 그를 굽어보는 측면.\n\nLOCATION (lock): At the boat-side edge of the mainland pier beside the crowded fishing vessel. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the crowded fishing-boat deck, sunny harbor, passengers, and pier beside it. Exclude the broad passenger emphasis and show the boatman speaking beside the officer.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (한국인 40대 중후반 남성, 호리한 얼굴, 짧은 검은 머리, 가무잡잡한 피부) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 8,
   "A": 5,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "선착장에 선 구창모와 그를 배 위에서 굽어보는 황봉업의 측면 구도를 정확히 구현했으며, 이전 샷의 승객 위치와 복장 연속성도 훌륭하게 유지했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "지문에서 요구한 '선착장의 구창모'와 달리 구창모가 배 위에 서 있으며, 상대를 굽어보는 구도(높낮이 차이)가 제대로 표현되지 않았습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "레퍼런스로 지정된 구창모 캐릭터가 완전히 누락되었으며, 이전 샷에서 배 중앙에 서 있던 인물을 선착장에 앉혀놓아 샷 연속성을 심각하게 훼손했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S101sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:893415>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "선착장에 서 있는 구창모가 레퍼런스 이미지의 경찰 반팔 셔츠 대신 이전 샷에 등장한 다른 인물의 검은색 다용도 조끼를 입고 있습니다.",
     "fix_en": "Change the clothing of the man standing on the pier to the short-sleeved blue police uniform shirt and duty belt shown in his character reference.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "배 위 승객 무리 중앙에 선착장에 있는 구창모와 동일한 얼굴에 검은색 조끼를 입은 복제된 인물이 서 있습니다.",
     "fix_en": "Remove the duplicate man in the black vest standing among the women on the boat, replacing him with the background of the boat.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "배 위에 서 있는 뱃사공(황봉업)의 복장이 이전 샷(왼쪽 끝 인물)의 남색 바람막이와 검은색 모자에서 파란색 패딩 조끼와 무늬 셔츠로 일관성 없이 변경되었습니다.",
     "fix_en": "Change the boatman's clothing to match the navy windbreaker, dark cap, and red life vest worn by the boatman on the far left of the previous still.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the clothing of the man standing on the pier to the short-sleeved blue police uniform shirt and duty belt shown in his character reference.\n- Remove the duplicate man in the black vest standing among the women on the boat, replacing him with the background of the boat.\n- Change the boatman's clothing to match the navy windbreaker, dark cap, and red life vest worn by the boatman on the far left of the previous still.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S101sh1"
  }
 },
 "S101sh9": {
  "input_fingerprint": "8940a83dccbf3569",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 구창모가 호기심과 탐욕으로 두 눈을 번쩍 뜬 찰나의 얼굴 클로즈업.\n\nLOCATION (lock): Beside the moored fishing boat on the open mainland pier. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the bright boat-and-pier setting and the continuing conversation positions. Exclude the crowd from the close frame and focus on the officer's greedy reaction.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (한국인 40대 중후반 남성, 호리한 얼굴, 짧은 검은 머리, 가무잡잡한 피부) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "이전 숏의 대화 구도를 유지하면서 호기심과 탐욕으로 두 눈을 번쩍 뜬 표정 연기를 가장 정확하게 포착했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "대화 구도는 잘 유지되었으나 탐욕이나 호기심보다는 경직되거나 당황한 표정에 가깝습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "정면을 응시하여 이전 숏에서 이어지는 대화 구도를 깨뜨렸고, 요구된 표정 연기도 부족합니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S101sh6_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:893415>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트에 명시되지 않은 인물이 배경의 배 위에 추가되었습니다.",
     "fix_en": "Remove the blurry person standing on the boat in the background, replacing them with the blue structure of the boat and the background scenery.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "경찰 제복 견장에 이전 샷(PREVIOUS SHOT STILL)에는 없는 흰색 테두리와 큰 별 문양이 임의로 추가되었습니다.",
     "fix_en": "Remove the white borders and large white emblems from the shoulder epaulets, changing them to plain dark navy to match the previous still.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the blurry person standing on the boat in the background, replacing them with the blue structure of the boat and the background scenery.\n- Remove the white borders and large white emblems from the shoulder epaulets, changing them to plain dark navy to match the previous still.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S101sh6"
  }
 },
 "groupbg::동굴_내부": {
  "input_fingerprint": "c5a08487904c1918",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "동굴_내부",
    "tags": [
     "S102sh6",
     "S51sh2",
     "S51sh7",
     "S71sh14",
     "S71sh18",
     "S71sh2",
     "S75sh2",
     "S94sh2",
     "S94sh7",
     "S96sh1",
     "S96sh4",
     "S98sh2",
     "S98sh5",
     "S98sh6"
    ]
   },
   "context_sig": "68a1c19012c015b2"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: At the chained holding area inside the dark cave chamber during daylight.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 숲속 동굴 내부: 빛이 들지 않는 깊고 축축한 자연 동굴. 바닥에 인골과 토막 난 시체가 널려 있다. (특징: 울퉁불퉁하고 어두운 암석 벽과 천장; 천장에 빽빽이 매달린 수백 마리 박쥐 떼; 바닥에 흩어진 사람의 뼈와 토막 난 신체 부위; 바닥에 고인 피 웅덩이; 벽에 고정된 무거운 쇠사슬) / 금월도 숲속 동굴 입구: 거대한 자연 동굴의 입구. 나무 기둥과 판자를 얼기설기 덧대어 형태를 만들었고 내부는 캄캄하다. (특징: 거대한 암석 동굴 구멍; 입구에 덧댄 낡은 나무 기둥과 판자; 캄캄하고 어두운 동굴 안쪽) / 금월도 해송 숲길 및 숲속: 해송이 울창하게 우거져 빛이 잘 들지 않는 빽빽하고 좁은 흙길과 숲의 심층부. (특징: 빽빽한 해송 나무 기둥들; 무성하고 어두운 수풀; 바닥에 깔린 흙과 솔잎; ‘수리영’이 적힌 이끼 낀 비석들; 비석 앞에 놓인 썩은 핏빛 육고기 덩어리) / 반월 식당 뒷방: 식당 건물 안쪽에 위치한 주거용 방. 정갈하게 정돈되어 햇살이 스며든다. (특징: 정돈된 방바닥과 이부자리; 격자무늬 나무 창; 창문 너머 들어오는 햇살; 옷걸이에 걸린 화려한 붉은색 무녀복) / 반월 식당 앞: 식당 건물 앞의 골목길 및 입구 흙길 지형. (특징: 미닫이 나무 문; 낡은 식당 외벽; 흙 덮인 골목길; 바닥의 짙은 핏자국) / 해안마을 선착장: 육지 쪽 항구의 선착장. 어선들이 정박해 있고 콘크리트 바닥에 폴리스 라인이 쳐져 있다. (특징: 콘크리트 항구 부두 바닥; 드럼통 모닥불; 시체 주변의 노란색 폴리스 라인; 구경꾼 무리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실내. 동굴 - 밤\n- 초췌한 몰골, 쇠사슬에 묶인 상태…\n- 실내. 동굴 안 - 밤\n- 실내. 동굴 안 - 밤\n- 실내. 숲 속 동굴 - 낮\n- 실내. 동굴 - 낮\n- 다시 동굴 - 낮\n- 실내. 동굴 - 낮\n\nTIME OF DAY (lock): day transitioning through dusk to evening.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: At the chained holding area inside the dark cave chamber during daylight.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 숲속 동굴 내부: 빛이 들지 않는 깊고 축축한 자연 동굴. 바닥에 인골과 토막 난 시체가 널려 있다. (특징: 울퉁불퉁하고 어두운 암석 벽과 천장; 천장에 빽빽이 매달린 수백 마리 박쥐 떼; 바닥에 흩어진 사람의 뼈와 토막 난 신체 부위; 바닥에 고인 피 웅덩이; 벽에 고정된 무거운 쇠사슬) / 금월도 숲속 동굴 입구: 거대한 자연 동굴의 입구. 나무 기둥과 판자를 얼기설기 덧대어 형태를 만들었고 내부는 캄캄하다. (특징: 거대한 암석 동굴 구멍; 입구에 덧댄 낡은 나무 기둥과 판자; 캄캄하고 어두운 동굴 안쪽) / 금월도 해송 숲길 및 숲속: 해송이 울창하게 우거져 빛이 잘 들지 않는 빽빽하고 좁은 흙길과 숲의 심층부. (특징: 빽빽한 해송 나무 기둥들; 무성하고 어두운 수풀; 바닥에 깔린 흙과 솔잎; ‘수리영’이 적힌 이끼 낀 비석들; 비석 앞에 놓인 썩은 핏빛 육고기 덩어리) / 반월 식당 뒷방: 식당 건물 안쪽에 위치한 주거용 방. 정갈하게 정돈되어 햇살이 스며든다. (특징: 정돈된 방바닥과 이부자리; 격자무늬 나무 창; 창문 너머 들어오는 햇살; 옷걸이에 걸린 화려한 붉은색 무녀복) / 반월 식당 앞: 식당 건물 앞의 골목길 및 입구 흙길 지형. (특징: 미닫이 나무 문; 낡은 식당 외벽; 흙 덮인 골목길; 바닥의 짙은 핏자국) / 해안마을 선착장: 육지 쪽 항구의 선착장. 어선들이 정박해 있고 콘크리트 바닥에 폴리스 라인이 쳐져 있다. (특징: 콘크리트 항구 부두 바닥; 드럼통 모닥불; 시체 주변의 노란색 폴리스 라인; 구경꾼 무리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실내. 동굴 - 밤\n- 초췌한 몰골, 쇠사슬에 묶인 상태…\n- 실내. 동굴 안 - 밤\n- 실내. 동굴 안 - 밤\n- 실내. 숲 속 동굴 - 낮\n- 실내. 동굴 - 낮\n- 다시 동굴 - 낮\n- 실내. 동굴 - 낮\n\nTIME OF DAY (lock): day transitioning through dusk to evening.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_동굴_내부_fbbdde.png",
  "asset_id": "a2cee75a-0d96-4fda-a37e-04d78b950d6e",
  "input_asset_ids": [
   "aef66cc9-fe18-495e-9efe-3f4dd20e9857"
  ],
  "origin_tag": "S102sh6",
  "place_text": "At the chained holding area inside the dark cave chamber during daylight.",
  "origin_inputs": {
   "place_text": "At the chained holding area inside the dark cave chamber during daylight.",
   "time_of_day_en": "day transitioning through dusk to evening",
   "conti_asset_id": "aef66cc9-fe18-495e-9efe-3f4dd20e9857"
  }
 },
 "S102sh6::bgfirst_bg": {
  "input_fingerprint": "188924df9ada3360",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 양손으로 굵은 쇠사슬을 꽉 움켜쥔 채 온몸을 뒤로 젖히며 팽팽하게 당기고 있는 김지후(강수리영)의 긴장된 자세.\n\nLOCATION (lock): At the chained holding area inside the dark cave chamber during daylight.\n\nTIME OF DAY (lock): day transitioning through dusk to evening.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: The lateral track from 정인우's side settles at waist height into a medium-full profile of 김지후(강수리영), preserving her backward lean and the chain's taut line within one readable plane. Her braced body occupies one side of the frame while both hands and a visible length of chain pull toward the opposite edge, making resistance physically clear without enlarging the prop unnaturally.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 굵은 쇠사슬 (taut and unmoving) — Its visible length runs taut from her hands toward the opposite frame edge; used as Creates the diagonal tension line from 김지후(강수리영)'s clenched hands toward its unseen restraint point; 동굴 내부 (daytime cave setting); used as Frames the pulling action and maintains continuity with the preceding cave exchange.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 양손으로 굵은 쇠사슬을 꽉 움켜쥔 채 온몸을 뒤로 젖히며 팽팽하게 당기고 있는 김지후(강수리영)의 긴장된 자세.\n\nLOCATION (lock): At the chained holding area inside the dark cave chamber during daylight.\n\nTIME OF DAY (lock): day transitioning through dusk to evening.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: The lateral track from 정인우's side settles at waist height into a medium-full profile of 김지후(강수리영), preserving her backward lean and the chain's taut line within one readable plane. Her braced body occupies one side of the frame while both hands and a visible length of chain pull toward the opposite edge, making resistance physically clear without enlarging the prop unnaturally.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 굵은 쇠사슬 (taut and unmoving) — Its visible length runs taut from her hands toward the opposite frame edge; used as Creates the diagonal tension line from 김지후(강수리영)'s clenched hands toward its unseen restraint point; 동굴 내부 (daytime cave setting); used as Frames the pulling action and maintains continuity with the preceding cave exchange.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S102sh6__bgfirst_bg.png",
  "asset_id": "0c2d2f99-36e1-4b2a-97f9-7bd01e3dc281",
  "input_asset_ids": [
   "aef66cc9-fe18-495e-9efe-3f4dd20e9857",
   "a2cee75a-0d96-4fda-a37e-04d78b950d6e"
  ]
 },
 "S102sh6": {
  "input_fingerprint": "7b2e8c186b15a909",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day transitioning through dusk to evening.\n\nSHOT TEXT (authoritative, Korean): 양손으로 굵은 쇠사슬을 꽉 움켜쥔 채 온몸을 뒤로 젖히며 팽팽하게 당기고 있는 김지후(강수리영)의 긴장된 자세.\n\nLOCATION (lock): At the chained holding area inside the dark cave chamber during daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is upright with both hands locked around the heavy chain, her arms pulled taut and her whole torso arched backward while the fixed chain prevents her from moving away.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu and Inwoo remain chained in the cave; Inwoo’s face remains injured and Jihu’s pendant remains with the ritual objects.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day transitioning through dusk to evening.\n\nSHOT TEXT (authoritative, Korean): 양손으로 굵은 쇠사슬을 꽉 움켜쥔 채 온몸을 뒤로 젖히며 팽팽하게 당기고 있는 김지후(강수리영)의 긴장된 자세.\n\nLOCATION (lock): At the chained holding area inside the dark cave chamber during daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is upright with both hands locked around the heavy chain, her arms pulled taut and her whole torso arched backward while the fixed chain prevents her from moving away.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu and Inwoo remain chained in the cave; Inwoo’s face remains injured and Jihu’s pendant remains with the ritual objects.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day transitioning through dusk to evening.\n\nSHOT TEXT (authoritative, Korean): 양손으로 굵은 쇠사슬을 꽉 움켜쥔 채 온몸을 뒤로 젖히며 팽팽하게 당기고 있는 김지후(강수리영)의 긴장된 자세.\n\nLOCATION (lock): At the chained holding area inside the dark cave chamber during daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is upright with both hands locked around the heavy chain, her arms pulled taut and her whole torso arched backward while the fixed chain prevents her from moving away.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu and Inwoo remain chained in the cave; Inwoo’s face remains injured and Jihu’s pendant remains with the ritual objects.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S102sh6__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S102sh6.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_동굴_내부_fbbdde.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "사슬을 쥐고 온몸을 뒤로 젖히는 자세와 부상당한 인우, 제단 위의 펜던트 등 모든 요구 사항을 충실히 반영했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지문에 명시되지 않은 세 번째 인물이 배경에 등장하여 '발명된 인물(Invented people)' 규정 위반으로 실격입니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "사슬을 쥐고 온몸을 뒤로 젖히는 자세와 부상당한 인우, 제단 위의 펜던트 등 모든 요구 사항을 충실히 반영했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지문에 명시되지 않은 세 번째 인물이 배경에 등장하여 '발명된 인물(Invented people)' 규정 위반으로 실격입니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "쇠사슬을 당기며 몸을 젖히는 긴장된 자세를 잘 구현했으며, 부상당한 인우와 제단 위의 펜던트 등 명시된 설정들을 정확히 반영했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "배경에 지시되지 않은 여분의 인물이 추가로 등장하여 불필요한 사람을 생성하면 안 된다는 핵심 규정(Hard Violation)을 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "쇠사슬을 당기며 몸을 젖히는 긴장된 자세를 잘 구현했으며, 부상당한 인우와 제단 위의 펜던트 등 명시된 설정들을 정확히 반영했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "배경에 지시되지 않은 여분의 인물이 추가로 등장하여 불필요한 사람을 생성하면 안 된다는 핵심 규정(Hard Violation)을 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "사슬을 쥐고 온몸을 뒤로 젖히는 자세와 부상당한 인우, 제단 위의 펜던트 등 모든 요구 사항을 충실히 반영했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "지문에 명시되지 않은 세 번째 인물이 배경에 등장하여 '발명된 인물(Invented people)' 규정 위반으로 실격입니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_동굴_내부_fbbdde.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지문에는 김지후만 명시되어 있으나, 배경에 다른 인물이 등장합니다.",
     "fix_en": "Remove the man sitting in the background, filling the space seamlessly with the cave wall and floor.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "캐릭터가 쇠사슬을 잡고 있는 손이 3개로 그려져 있습니다.",
     "fix_en": "Remove the extra third hand from the chain, leaving only the two hands that correctly attach to her left and right arms.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the man sitting in the background, filling the space seamlessly with the cave wall and floor.\n- Remove the extra third hand from the chain, leaving only the two hands that correctly attach to her left and right arms.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S102sh6__bgfirst_bg.png",
   "bg_asset_id": "0c2d2f99-36e1-4b2a-97f9-7bd01e3dc281",
   "bg_record_key": "S102sh6::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "동굴_내부",
   "groupbg_asset_id": "a2cee75a-0d96-4fda-a37e-04d78b950d6e"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S102sh12::bgfirst_bg": {
  "input_fingerprint": "72cd247735512eec",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 열린 어창 바닥에 손발이 묶이고 입이 천으로 틀어막힌 어창에 감금된 여성 제물들이 공포에 질려 잔뜩 웅크린 상태.\n\nLOCATION (lock): Inside the open fish hold beneath a fishing boat's deck, where bound captives are packed together.\n\nTIME OF DAY (lock): day transitioning through dusk to evening.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From above the opened boat hold, a steep high angle advances into a tight wide composition that keeps several captive bodies, their bound limbs, and covered mouths simultaneously legible. The women cluster across the lower and middle frame in overlapping curled poses, while the rectangular hold opening remains visible around them as the scale reference and enclosure.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: open boat hold rim in the middle-center of the frame, foreground; clustered captive women in the lower-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 어창 입구 (open) — The opening faces upward toward the camera, exposing the hold interior below; used as Forms a visible enclosure around the clustered captives and guides the steep downward composition; 어창 뚜껑 (opened) — The raised lids expose their inner-facing sides around the open compartments; used as Visible around the opening as evidence that the compartment has been uncovered; 손발의 결박과 입을 막은 천 (fastened in place); used as Makes the women's captivity immediately readable within the group framing.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 열린 어창 바닥에 손발이 묶이고 입이 천으로 틀어막힌 어창에 감금된 여성 제물들이 공포에 질려 잔뜩 웅크린 상태.\n\nLOCATION (lock): Inside the open fish hold beneath a fishing boat's deck, where bound captives are packed together.\n\nTIME OF DAY (lock): day transitioning through dusk to evening.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From above the opened boat hold, a steep high angle advances into a tight wide composition that keeps several captive bodies, their bound limbs, and covered mouths simultaneously legible. The women cluster across the lower and middle frame in overlapping curled poses, while the rectangular hold opening remains visible around them as the scale reference and enclosure.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: open boat hold rim in the middle-center of the frame, foreground; clustered captive women in the lower-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 어창 입구 (open) — The opening faces upward toward the camera, exposing the hold interior below; used as Forms a visible enclosure around the clustered captives and guides the steep downward composition; 어창 뚜껑 (opened) — The raised lids expose their inner-facing sides around the open compartments; used as Visible around the opening as evidence that the compartment has been uncovered; 손발의 결박과 입을 막은 천 (fastened in place); used as Makes the women's captivity immediately readable within the group framing.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S102sh12__bgfirst_bg.png",
  "asset_id": "b530e4e6-c38e-4664-93c7-5d7d10170fb3",
  "input_asset_ids": [
   "78cd6e19-b6a4-4a7a-ae1a-f7897784aaac",
   "81ef185c-4c25-41b8-ad18-10745e6f506f"
  ]
 },
 "S102sh12": {
  "input_fingerprint": "ef5dfcdce73679a0",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day transitioning through dusk to evening.\n\nSHOT TEXT (authoritative, Korean): 열린 어창 바닥에 손발이 묶이고 입이 천으로 틀어막힌 어창에 감금된 여성 제물들이 공포에 질려 잔뜩 웅크린 상태.\n\nLOCATION (lock): Inside the open fish hold beneath a fishing boat's deck, where bound captives are packed together. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): The women are packed tightly on the floor of the open fish holds with their wrists and ankles bound and their mouths gagged, their torsos hunched and their knees drawn inward in fear.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The women hidden in the fish holds remain bound hand and foot and gagged, unable to move freely.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day transitioning through dusk to evening.\n\nSHOT TEXT (authoritative, Korean): 열린 어창 바닥에 손발이 묶이고 입이 천으로 틀어막힌 어창에 감금된 여성 제물들이 공포에 질려 잔뜩 웅크린 상태.\n\nLOCATION (lock): Inside the open fish hold beneath a fishing boat's deck, where bound captives are packed together. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): The women are packed tightly on the floor of the open fish holds with their wrists and ankles bound and their mouths gagged, their torsos hunched and their knees drawn inward in fear.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The women hidden in the fish holds remain bound hand and foot and gagged, unable to move freely.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day transitioning through dusk to evening.\n\nSHOT TEXT (authoritative, Korean): 열린 어창 바닥에 손발이 묶이고 입이 천으로 틀어막힌 어창에 감금된 여성 제물들이 공포에 질려 잔뜩 웅크린 상태.\n\nLOCATION (lock): Inside the open fish hold beneath a fishing boat's deck, where bound captives are packed together. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): The women are packed tightly on the floor of the open fish holds with their wrists and ankles bound and their mouths gagged, their torsos hunched and their knees drawn inward in fear.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The women hidden in the fish holds remain bound hand and foot and gagged, unable to move freely.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S102sh12__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S102sh12.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_mainland_harbor_dock_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "레퍼런스의 항구 로케이션 배경과 해질녘 시간대를 정확히 반영했으며, 어창 내 인물들의 결박 및 웅크린 포즈도 지시사항을 잘 따랐습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지정된 로케이션의 배경 구조물이 보이지 않고 시간대도 맞지 않으며, 어창이 어선 내부가 아닌 콘크리트 부두에 있는 것처럼 묘사되었습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "레퍼런스의 항구 로케이션 배경과 해질녘 시간대를 정확히 반영했으며, 어창 내 인물들의 결박 및 웅크린 포즈도 지시사항을 잘 따랐습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "지정된 로케이션의 배경 구조물이 보이지 않고 시간대도 맞지 않으며, 어창이 어선 내부가 아닌 콘크리트 부두에 있는 것처럼 묘사되었습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 해 질 녘 시간대와 결박된 채 웅크린 여성들의 모습, 그리고 레퍼런스 이미지의 방파제와 마을 배경을 정확히 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "시간대가 밝은 낮으로 잘못 묘사되었으며, 레퍼런스의 배경 디테일이 생략되었고 갇힌 공간이 어창보다는 콘크리트 박스처럼 보입니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지정된 해 질 녘 시간대와 결박된 채 웅크린 여성들의 모습, 그리고 레퍼런스 이미지의 방파제와 마을 배경을 정확히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "시간대가 밝은 낮으로 잘못 묘사되었으며, 레퍼런스의 배경 디테일이 생략되었고 갇힌 공간이 어창보다는 콘크리트 박스처럼 보입니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "레퍼런스의 항구 로케이션 배경과 해질녘 시간대를 정확히 반영했으며, 어창 내 인물들의 결박 및 웅크린 포즈도 지시사항을 잘 따랐습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "지정된 로케이션의 배경 구조물이 보이지 않고 시간대도 맞지 않으며, 어창이 어선 내부가 아닌 콘크리트 부두에 있는 것처럼 묘사되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_mainland_harbor_dock_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "화면 오른쪽 앞줄에 앉은 두 여성의 노출된 발목에 밧줄이 없어, 모든 여성의 발목이 묶여 있어야 한다는 지시를 위반했습니다.",
     "fix_en": "Add tight rope bindings around the visible bare ankles of the two women sitting on the right side.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add tight rope bindings around the visible bare ankles of the two women sitting on the right side.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S102sh12__bgfirst_bg.png",
   "bg_asset_id": "b530e4e6-c38e-4664-93c7-5d7d10170fb3",
   "bg_record_key": "S102sh12::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S102sh17::bgfirst_bg": {
  "input_fingerprint": "d3ac03405f991519",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 순백의 무녀복을 입은 강영란(흰 머리 여자)이 제단 앞 붉은 모닥불 불빛을 받으며 꼿꼿하고 위엄 있게 선 전신.\n\nLOCATION (lock): Before the boulder altar in the evening ritual clearing, lit by red campfires.\n\nTIME OF DAY (lock): day transitioning through dusk to evening.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From waist height four meters in front of the altar, the camera completes its central-axis dolly and stops with 강영란(흰 머리 여자) framed head to toe in the middle third. A slight low-angle tilt preserves her rigid ceremonial authority, while the altar and worshippers remain layered behind her and the fire stays below the foreground-object occupancy limit.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: altar before the rock (Set for the Full Moon Festival ritual) — Its front side faces the camera on the same central axis as 강영란(흰 머리 여자); used as Forms the symmetrical background anchor directly behind the ritual figure; bonfire (Burning strongly); used as Provides the documented red illumination and a low foreground counterweight; circle of black-clad residents (Gathered around the altar in prayer); used as Creates a dark peripheral enclosure around the centered white-clad figure.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 순백의 무녀복을 입은 강영란(흰 머리 여자)이 제단 앞 붉은 모닥불 불빛을 받으며 꼿꼿하고 위엄 있게 선 전신.\n\nLOCATION (lock): Before the boulder altar in the evening ritual clearing, lit by red campfires.\n\nTIME OF DAY (lock): day transitioning through dusk to evening.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From waist height four meters in front of the altar, the camera completes its central-axis dolly and stops with 강영란(흰 머리 여자) framed head to toe in the middle third. A slight low-angle tilt preserves her rigid ceremonial authority, while the altar and worshippers remain layered behind her and the fire stays below the foreground-object occupancy limit.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: altar before the rock (Set for the Full Moon Festival ritual) — Its front side faces the camera on the same central axis as 강영란(흰 머리 여자); used as Forms the symmetrical background anchor directly behind the ritual figure; bonfire (Burning strongly); used as Provides the documented red illumination and a low foreground counterweight; circle of black-clad residents (Gathered around the altar in prayer); used as Creates a dark peripheral enclosure around the centered white-clad figure.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S102sh17__bgfirst_bg.png",
  "asset_id": "45926bfc-933a-426e-9c87-8d0d4ac933f1",
  "input_asset_ids": [
   "2c3e612b-c736-490c-a424-3eede267f32e",
   "81ef185c-4c25-41b8-ad18-10745e6f506f"
  ]
 },
 "S102sh17": {
  "input_fingerprint": "7b5b7422076ef906",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day transitioning through dusk to evening.\n\nSHOT TEXT (authoritative, Korean): 순백의 무녀복을 입은 강영란(흰 머리 여자)이 제단 앞 붉은 모닥불 불빛을 받으며 꼿꼿하고 위엄 있게 선 전신.\n\nLOCATION (lock): Before the boulder altar in the evening ritual clearing, lit by red campfires. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran has changed from the red robe into a pure white shaman’s robe; Jihu’s pendant remains among her ritual objects.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day transitioning through dusk to evening.\n\nSHOT TEXT (authoritative, Korean): 순백의 무녀복을 입은 강영란(흰 머리 여자)이 제단 앞 붉은 모닥불 불빛을 받으며 꼿꼿하고 위엄 있게 선 전신.\n\nLOCATION (lock): Before the boulder altar in the evening ritual clearing, lit by red campfires. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran has changed from the red robe into a pure white shaman’s robe; Jihu’s pendant remains among her ritual objects.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day transitioning through dusk to evening.\n\nSHOT TEXT (authoritative, Korean): 순백의 무녀복을 입은 강영란(흰 머리 여자)이 제단 앞 붉은 모닥불 불빛을 받으며 꼿꼿하고 위엄 있게 선 전신.\n\nLOCATION (lock): Before the boulder altar in the evening ritual clearing, lit by red campfires. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran has changed from the red robe into a pure white shaman’s robe; Jihu’s pendant remains among her ritual objects.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S102sh17__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S102sh17.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_mainland_harbor_dock_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 로케이션(콘크리트 부두)을 정확히 구현했으며, 프롬프트에 명시된 인물만 정확한 조명과 의상으로 연출했습니다."
     },
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "프롬프트에 언급되지 않은 다수의 인물이 등장하는 치명적인 위반이 있으며, 지정된 로케이션의 바닥 재질(콘크리트)을 흙바닥으로 임의 변경했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 로케이션(콘크리트 부두)을 정확히 구현했으며, 프롬프트에 명시된 인물만 정확한 조명과 의상으로 연출했습니다."
     },
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "프롬프트에 언급되지 않은 다수의 인물이 등장하는 치명적인 위반이 있으며, 지정된 로케이션의 바닥 재질(콘크리트)을 흙바닥으로 임의 변경했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지정된 로케이션(콘크리트 부두)과 등장인물 제한(추가 인물 없음)을 정확히 준수하며 전신 샷의 요구사항을 잘 구현했습니다."
     },
     {
      "label": "B",
      "score": 2,
      "verdict_ko": "지정되지 않은 다수의 인물을 추가로 배치했고(심각한 위반), 로케이션 레퍼런스의 구조(콘크리트 바닥 및 배경)를 무시했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 로케이션(콘크리트 부두)과 등장인물 제한(추가 인물 없음)을 정확히 준수하며 전신 샷의 요구사항을 잘 구현했습니다."
     },
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "지정되지 않은 다수의 인물을 추가로 배치했고(심각한 위반), 로케이션 레퍼런스의 구조(콘크리트 바닥 및 배경)를 무시했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 4,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 4,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "지정된 로케이션(콘크리트 부두)을 정확히 구현했으며, 프롬프트에 명시된 인물만 정확한 조명과 의상으로 연출했습니다."
   },
   {
    "label": "A",
    "score": 2,
    "verdict_ko": "프롬프트에 언급되지 않은 다수의 인물이 등장하는 치명적인 위반이 있으며, 지정된 로케이션의 바닥 재질(콘크리트)을 흙바닥으로 임의 변경했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_mainland_harbor_dock_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "여성의 양손 묘사가 해부학적으로 어색합니다 (화면 좌측 손가락은 뭉개졌고, 우측 손은 소매 끝에 보이지 않음).",
     "fix_en": "Restore anatomically correct hands with defined fingers emerging from the sleeves on both sides, resting naturally.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "배경의 작은 콘크리트 건물 정면이 기준 사진(좌측 문 1개, 우측 창문 1개)과 달리 두 개의 문으로 잘못 그려졌습니다.",
     "fix_en": "Change the right-side door of the small concrete building in the background into a square window to match the reference structure.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "화면 좌측의 경사면이 기준 사진의 자연스러운 흙과 풀밭이 아닌 인공적인 돌담(축대)으로 잘못 변형되었습니다.",
     "fix_en": "Replace the stone retaining wall on the left with a natural dirt and dry grass embankment matching the reference photograph.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "바위 제단 위에 놓인 의식용 물품들 사이에 프롬프트에서 요구한 펜던트가 보이지 않습니다.",
     "fix_en": "Add a clearly visible pendant among the ritual objects resting on the boulder altar.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore anatomically correct hands with defined fingers emerging from the sleeves on both sides, resting naturally.\n- Change the right-side door of the small concrete building in the background into a square window to match the reference structure.\n- Replace the stone retaining wall on the left with a natural dirt and dry grass embankment matching the reference photograph.\n- Add a clearly visible pendant among the ritual objects resting on the boulder altar.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S102sh17__bgfirst_bg.png",
   "bg_asset_id": "45926bfc-933a-426e-9c87-8d0d4ac933f1",
   "bg_record_key": "S102sh17::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S103sh2::bgfirst_bg": {
  "input_fingerprint": "5c2ee999cff4278a",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 바닷물에 흠뻑 젖은 고산들이 갯바위 위에 대자로 벌러덩 드러누워 있는 전신.\n\nLOCATION (lock): On wave-washed coastal rocks at evening, immediately above the sea.\n\nTIME OF DAY (lock): evening, fading light.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From high above and slightly beyond 고산들's feet, the camera begins its diagonal advance along his body axis, holding his entire spread-eagle figure against the 갯바위. His body occupies the central vertical corridor, with the wave-struck edge beyond his feet providing scale and danger without obscuring him.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 갯바위 (Waves strike its edge) — The upper plane faces the high camera, with its sea-facing edge visible beyond 고산들's feet; used as Supports the full-body silhouette and defines the elevated camera geometry; sea (Moving in waves against the rock); used as Borders the rock and reinforces 고산들's recent emergence from the sea.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 바닷물에 흠뻑 젖은 고산들이 갯바위 위에 대자로 벌러덩 드러누워 있는 전신.\n\nLOCATION (lock): On wave-washed coastal rocks at evening, immediately above the sea.\n\nTIME OF DAY (lock): evening, fading light.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From high above and slightly beyond 고산들's feet, the camera begins its diagonal advance along his body axis, holding his entire spread-eagle figure against the 갯바위. His body occupies the central vertical corridor, with the wave-struck edge beyond his feet providing scale and danger without obscuring him.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 갯바위 (Waves strike its edge) — The upper plane faces the high camera, with its sea-facing edge visible beyond 고산들's feet; used as Supports the full-body silhouette and defines the elevated camera geometry; sea (Moving in waves against the rock); used as Borders the rock and reinforces 고산들's recent emergence from the sea.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S103sh2__bgfirst_bg.png",
  "asset_id": "94cef76b-981d-4ac0-8a6b-dffd1ff93e18",
  "input_asset_ids": [
   "b0372a1b-e2eb-4f10-9476-7f3027282984",
   "33cc6c13-db9b-4db6-bd03-c0c362696e4d"
  ]
 },
 "S103sh2": {
  "input_fingerprint": "1f6fa9b6ea9f553b",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): evening, fading light.\n\nSHOT TEXT (authoritative, Korean): 바닷물에 흠뻑 젖은 고산들이 갯바위 위에 대자로 벌러덩 드러누워 있는 전신.\n\nLOCATION (lock): On wave-washed coastal rocks at evening, immediately above the sea. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul remains soaked from the sea and bears the deep back-and-chest stab wounds and earlier shoulder wound.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): evening, fading light.\n\nSHOT TEXT (authoritative, Korean): 바닷물에 흠뻑 젖은 고산들이 갯바위 위에 대자로 벌러덩 드러누워 있는 전신.\n\nLOCATION (lock): On wave-washed coastal rocks at evening, immediately above the sea. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul remains soaked from the sea and bears the deep back-and-chest stab wounds and earlier shoulder wound.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): evening, fading light.\n\nSHOT TEXT (authoritative, Korean): 바닷물에 흠뻑 젖은 고산들이 갯바위 위에 대자로 벌러덩 드러누워 있는 전신.\n\nLOCATION (lock): On wave-washed coastal rocks at evening, immediately above the sea. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul remains soaked from the sea and bears the deep back-and-chest stab wounds and earlier shoulder wound.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S103sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S103sh2.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L15B03.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "'대자로 벌러덩 드러누워 있는 전신'이라는 지문에 완벽하게 부합하도록 팔다리를 넓게 벌린 자세를 훌륭하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "장소와 전신 샷의 조건은 충족했으나, 다리가 모아져 있어 '대자로' 누운 특유의 자세를 충분히 살리지 못했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "'대자로 벌러덩 드러누워 있는 전신'이라는 지문에 완벽하게 부합하도록 팔다리를 넓게 벌린 자세를 훌륭하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "장소와 전신 샷의 조건은 충족했으나, 다리가 모아져 있어 '대자로' 누운 특유의 자세를 충분히 살리지 못했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지문이 요구한 '대자로 벌러덩 드러누워 있는' 자세를 팔다리를 넓게 벌려 완벽하게 구현했으며, 전신 앵글과 젖은 상태 및 상처 묘사도 우수합니다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "전신 샷과 배경 조건은 잘 충족하였으나, 다리를 모으고 누워 있어 '대자'로 누워 있다는 가장 중요한 핵심 자세 지문을 살리지 못했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지문이 요구한 '대자로 벌러덩 드러누워 있는' 자세를 팔다리를 넓게 벌려 완벽하게 구현했으며, 전신 앵글과 젖은 상태 및 상처 묘사도 우수합니다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "전신 샷과 배경 조건은 잘 충족하였으나, 다리를 모으고 누워 있어 '대자'로 누워 있다는 가장 중요한 핵심 자세 지문을 살리지 못했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 11
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 15,
   "B": 11
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "'대자로 벌러덩 드러누워 있는 전신'이라는 지문에 완벽하게 부합하도록 팔다리를 넓게 벌린 자세를 훌륭하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "장소와 전신 샷의 조건은 충족했으나, 다리가 모아져 있어 '대자로' 누운 특유의 자세를 충분히 살리지 못했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L15B03.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "화면 우측에 있는 인물의 왼쪽 발이 오른발처럼 잘못 그려져 있습니다(엄지발가락이 바깥쪽에 위치함).",
     "fix_en": "Redraw the character's left foot (on the right side of the image) to correctly depict a left foot with the big toe on the inside edge.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "유지 조건에 명시된 '이전에 입은 어깨 상처(earlier shoulder wound)'가 묘사되지 않았습니다.",
     "fix_en": "Add a visible wound and blood stain to the character's shoulder area.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the character's left foot (on the right side of the image) to correctly depict a left foot with the big toe on the inside edge.\n- Add a visible wound and blood stain to the character's shoulder area.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S103sh2__bgfirst_bg.png",
   "bg_asset_id": "94cef76b-981d-4ac0-8a6b-dffd1ff93e18",
   "bg_record_key": "S103sh2::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S103sh3": {
  "input_fingerprint": "3d23d79dbc32b788",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): evening, fading light.\n\nSHOT TEXT (authoritative, Korean): 갯바위에 누운 고산들이 목에 핏대를 세우고 허공을 향해 분노에 찬 괴성을 내지르는 찰나.\n\nLOCATION (lock): On the exposed coastal rocks where the soaked man has collapsed after climbing from the water. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the wet rocks, evening sea, soaked clothing, and the man's sprawled position. Exclude the exhausted stillness and show him crying out in rage.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul’s clothes remain soaked, and his accumulated stab and gunshot wounds remain visible as he cries out.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "레퍼런스의 누운 자세와 열린 셔츠 속 상처를 정확히 유지하면서, 목에 핏대를 세우고 포효하는 지문을 훌륭하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "누운 자세는 유지했으나 셔츠가 잠겨 있어 유지되어야 할 상처가 가려졌으며, 괴성을 지르는 묘사도 비교적 약합니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "이전 샷의 누운 자세와 접촉면을 그대로 유지해야 한다는 지침을 무시하고 상체를 일으켰으며 상처도 보이지 않습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S103sh2_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "화면 왼쪽(남자의 오른손)을 보면, 팔 안쪽이 위를 향하고 있어 손바닥이 하늘을 보는 자세임에도 엄지손가락이 발 쪽을 향해 달려 있어 해부학적으로 틀렸습니다.",
     "fix_en": "Redraw the man's right hand (viewer's left) so the thumb points up towards his head, matching the natural anatomy of a supinated arm.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "참조 이미지에서는 남자의 오른쪽 셔츠(화면 왼쪽)가 비교적 깨끗하지만, 생성된 이미지에는 셔츠 전체를 적시는 거대한 핏자국이 추가되어 이전 샷의 상태가 유지되어야 한다는 지시를 위반했습니다.",
     "fix_en": "Remove the heavy blood stains from the right side of the man's shirt (viewer's left) and restore the original, smaller chest wound shapes to match the reference image.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the man's right hand (viewer's left) so the thumb points up towards his head, matching the natural anatomy of a supinated arm.\n- Remove the heavy blood stains from the right side of the man's shirt (viewer's left) and restore the original, smaller chest wound shapes to match the reference image.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S103sh2"
  }
 },
 "S104sh1::bgfirst_bg": {
  "input_fingerprint": "8183cd37a5d012ff",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 식당 문턱을 넘는 자세로 양손에 짐꾸러미를 든 채, 경악하여 두 눈이 휘둥그레진 황봉업(황 씨)의 얼굴 클로즈업.\n\nLOCATION (lock): At the island restaurant threshold at night, stepping from the interior toward the front lane.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: The static camera waits outside the doorway at 황봉업(황 씨)'s eye level, compressing his face and shoulders into roughly the central two-thirds as he freezes halfway across the threshold. His head turns into a three-quarter angle toward the unseen road, while the two bundles enter only at the lower corners and the open doorway brackets the shock.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 황봉업(황 씨) in the middle-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: restaurant doorway (Open) — The camera faces the exterior side across the threshold, with the open passage directly behind 황봉업(황 씨); used as Frames the face on both sides and marks the interrupted exit; 짐꾸러미 (Held one in each hand); used as Appears partially at the lower corners as evidence that he was leaving.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 식당 문턱을 넘는 자세로 양손에 짐꾸러미를 든 채, 경악하여 두 눈이 휘둥그레진 황봉업(황 씨)의 얼굴 클로즈업.\n\nLOCATION (lock): At the island restaurant threshold at night, stepping from the interior toward the front lane.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: The static camera waits outside the doorway at 황봉업(황 씨)'s eye level, compressing his face and shoulders into roughly the central two-thirds as he freezes halfway across the threshold. His head turns into a three-quarter angle toward the unseen road, while the two bundles enter only at the lower corners and the open doorway brackets the shock.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 황봉업(황 씨) in the middle-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: restaurant doorway (Open) — The camera faces the exterior side across the threshold, with the open passage directly behind 황봉업(황 씨); used as Frames the face on both sides and marks the interrupted exit; 짐꾸러미 (Held one in each hand); used as Appears partially at the lower corners as evidence that he was leaving.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S104sh1__bgfirst_bg.png",
  "asset_id": "448896d5-bdb8-49d3-90d4-a9ec39226144",
  "input_asset_ids": [
   "8f934608-b721-456b-a147-daae6d369588",
   "e32f16fe-5ff4-40ac-9b6a-29c12cf48bea"
  ]
 },
 "S104sh1": {
  "input_fingerprint": "5302124daaab63cc",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 식당 문턱을 넘는 자세로 양손에 짐꾸러미를 든 채, 경악하여 두 눈이 휘둥그레진 황봉업(황 씨)의 얼굴 클로즈업.\n\nLOCATION (lock): At the island restaurant threshold at night, stepping from the interior toward the front lane. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul approaches with his wet clothing and accumulated stab and gunshot wounds unchanged.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 식당 문턱을 넘는 자세로 양손에 짐꾸러미를 든 채, 경악하여 두 눈이 휘둥그레진 황봉업(황 씨)의 얼굴 클로즈업.\n\nLOCATION (lock): At the island restaurant threshold at night, stepping from the interior toward the front lane. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul approaches with his wet clothing and accumulated stab and gunshot wounds unchanged.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 식당 문턱을 넘는 자세로 양손에 짐꾸러미를 든 채, 경악하여 두 눈이 휘둥그레진 황봉업(황 씨)의 얼굴 클로즈업.\n\nLOCATION (lock): At the island restaurant threshold at night, stepping from the interior toward the front lane. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul approaches with his wet clothing and accumulated stab and gunshot wounds unchanged.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S104sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S104sh1.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B05.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시된 얼굴 클로즈업 프레이밍과 경악하여 휘둥그레진 눈의 표정을 정확히 구현했으며 텍스트에 없는 인물을 배제했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "클로즈업 지시를 무시하고 프레임을 넓혔으며, 샷 텍스트에 명시되지 않은 인물을 배경에 임의로 추가하여 감점되었습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시된 얼굴 클로즈업 프레이밍과 경악하여 휘둥그레진 눈의 표정을 정확히 구현했으며 텍스트에 없는 인물을 배제했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "클로즈업 지시를 무시하고 프레임을 넓혔으며, 샷 텍스트에 명시되지 않은 인물을 배경에 임의로 추가하여 감점되었습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "지시된 경악하는 표정과 양손의 짐꾸러미를 잘 묘사했으며, 샷 텍스트에 없는 인물을 정확히 배제했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "샷 텍스트에 명시되지 않은 인물이 배경에 등장하여, 명시되지 않은 인물은 등장할 수 없다는 규칙을 위반했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지시된 경악하는 표정과 양손의 짐꾸러미를 잘 묘사했으며, 샷 텍스트에 없는 인물을 정확히 배제했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "샷 텍스트에 명시되지 않은 인물이 배경에 등장하여, 명시되지 않은 인물은 등장할 수 없다는 규칙을 위반했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 7
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 14,
   "B": 7
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지시된 얼굴 클로즈업 프레이밍과 경악하여 휘둥그레진 눈의 표정을 정확히 구현했으며 텍스트에 없는 인물을 배제했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "클로즈업 지시를 무시하고 프레임을 넓혔으며, 샷 텍스트에 명시되지 않은 인물을 배경에 임의로 추가하여 감점되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L35B05.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트에 명시된 젖은 옷과 누적된 자상 및 총상 흔적이 캐릭터에 반영되지 않았습니다.",
     "fix_en": "Apply a wet appearance to the character's clothing and add visible stab and gunshot wounds with blood to his chest and shirt.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Apply a wet appearance to the character's clothing and add visible stab and gunshot wounds with blood to his chest and shirt.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S104sh1__bgfirst_bg.png",
   "bg_asset_id": "448896d5-bdb8-49d3-90d4-a9ec39226144",
   "bg_record_key": "S104sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S104sh3": {
  "input_fingerprint": "a0a4afcdb1b21dc8",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 마을길 아래에서 살기 어린 눈빛을 한 고산들이 한쪽 발을 높이 들고 다른 발로 땅을 딛은 mid-stride 자세로 성큼성큼 걸어 올라오는 중인 위압적인 전신.\n\nLOCATION (lock): On the dark village lane climbing toward the island restaurant. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the restaurant frontage, dark uphill village road, doorway, and night lighting. Exclude the boatman's doorway close-up and show the soaked man advancing from below.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul remains soaked and wounded as he advances toward Hwang and Changmo.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 3,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "요구된 전신 구도와 걷는 자세를 잘 구현했으며, 텍스트에 없는 불필요한 인물을 추가하지 않고 프롬프트를 충실히 따랐습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "배경 식당 내부에 가방을 든 동일 인물이 중복으로 생성되는 치명적인 오류(Hard Violation)가 발생했습니다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "지시문에 명시되지 않은 정체불명의 인물들이 식당 입구에 등장하여 추가 인물 금지 규칙을 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S104sh1_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "레퍼런스 이미지에서 인물이 양손에 들고 있던 두 개의 천 보따리가 누락되었습니다.",
     "fix_en": "Place the two large, tied cloth bags from the reference image into the character's hands."
    },
    {
     "issue_ko": "프롬프트에 명시된 '한쪽 발을 높이 든' 역동적인 자세가 아니라 평범하게 걷고 있는 자세입니다.",
     "fix_en": "Raise the character's forward foot significantly higher into an exaggerated, dramatic mid-stride step."
    },
    {
     "issue_ko": "건물 좌측 상단의 파란색 주소 표지판에 텍스트와 숫자가 나타나 있습니다.",
     "fix_en": "Remove all text and numbers from the blue plaque on the upper left wall."
    },
    {
     "issue_ko": "인물이 어두운 마을길 아래에서 위로 걸어 올라와야 하는데, 식당 앞 평지에서 카메라를 향해(또는 내리막으로) 걸어오고 있어 위치와 방향이 잘못되었습니다.",
     "fix_en": "Relocating the character to the sloped path on the right and reversing his walking direction to face uphill cannot be done without altering the scene's geometry and viewpoint.",
     "needs_regeneration": true,
     "unfixable": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Place the two large, tied cloth bags from the reference image into the character's hands.\n- Raise the character's forward foot significantly higher into an exaggerated, dramatic mid-stride step.\n- Remove all text and numbers from the blue plaque on the upper left wall.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S104sh1"
  }
 },
 "S104sh7": {
  "input_fingerprint": "e32d60a5d23de15e",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 고산들이 도망치는 황봉업(황 씨)과 구창모의 등 뒤를 향해 맹수처럼 허공으로 맹렬하게 몸을 날린 순간.\n\nLOCATION (lock): Immediately outside the restaurant entrance, between the lane and the fleeing men at the doorway. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the restaurant entrance, dark road, and fleeing men's positions. Exclude the slow approach and show the pursuer leaping toward their backs.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul remains soaked and wounded as he leaps after the fleeing men.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 3,
   "C": 2
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "도망치는 두 사람의 등 뒤를 향해 허공으로 몸을 날리는 핵심 샷 연출을 지시대로 정확히 구현했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "인물들이 카메라를 등지고 있어 앞을 향해 뛰어오르는 구도가 되었으며, 레퍼런스의 피 묻은 의상이 다른 인물에게 중복 생성되었습니다."
   },
   {
    "label": "C",
    "score": 2,
    "verdict_ko": "레퍼런스 인물을 정지 상태로 복제해두고 도약하는 인물을 새로 만들어 초과 인물이 발생했습니다 (Hard violation)."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S104sh3_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "도약하는 남자가 레퍼런스 이미지에서 들고 있던 두 개의 하얀 비닐봉지를 쥐고 있지 않습니다.",
     "fix_en": "Add the two white bags from the reference image into the leaping man's hands."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the two white bags from the reference image into the leaping man's hands.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S104sh3"
  }
 },
 "S105sh1": {
  "input_fingerprint": "ee47539d294b6816",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, full moon entering eclipse.\n\nSHOT TEXT (authoritative, Korean): 둥근 보름달 빛이 스며드는 어두운 숲속 제단 한가운데에서 거대한 모닥불이 맹렬하게 불타오르는 전경.\n\nLOCATION (lock): At the center of the nighttime forest ritual clearing, before the boulder altar beneath the full moon. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran remains in the white shaman’s robe, and Jihu’s antler-bone pendant remains at the ritual altar.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "제단 중앙의 맹렬한 모닥불과 제단 위에 놓인 뿔 펜던트, 월식 중인 달의 모습을 훌륭하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "분위기와 배경은 잘 맞으나, 펜던트 대신 동물 두개골과 뼈가 제단에 배치되어 소품 지시를 어겼습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "제단의 형태가 명확하지 않고 뿔 소품이 흙바닥에 떨어져 있어 디테일과 구조적 요구사항을 놓쳤습니다."
   }
  ],
  "refs": [],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "lane(map_marker): 스케치",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S105sh3": {
  "input_fingerprint": "149f34401420ac5d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, full moon entering eclipse.\n\nSHOT TEXT (authoritative, Korean): 순백의 무녀복을 입은 강영란(흰 머리 여자)이 검붉은 피가 맺힌 잘린 염소 머리를 허공 높이 치켜든 채 정지된 찰나.\n\nLOCATION (lock): Directly before the boulder altar amid the blazing ritual fires. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the blazing bonfires, giant boulder, worshippers, full moon, and ritual smoke. Exclude the empty establishing emphasis and show the white-robed elder raising the severed goat head.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The sacrificial black goats remain tied before the rock as Youngran holds one severed goat head; her white robe begins to be stained with its blood.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (한국인 60대 후반 여성, 고운 노년 얼굴, 가지런히 틀어 묶은 머리, 기품 있는 이목구비); 검은 염소 (검은 털, 길쭉한 얼굴, 휘어진 두 뿔, 수평형 동공) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "A": 4,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "염소 머리와 흰 옷에 묻은 핏자국, 제단 앞에 묶여 있는 염소 등 프롬프트의 핵심 요소들을 가장 충실하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "염소를 치켜든 자세는 좋으나, 머리와 흰 옷에 있어야 할 필수적인 핏자국 묘사가 완전히 누락되었습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "핏자국이 전혀 없으며, 묶여 있어야 할 염소들이 끈 없이 서서 돌아다니고 화로의 위치도 레퍼런스와 다릅니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S105sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:831102>"
   },
   {
    "label": "CHARACTER REFERENCE — 검은 염소: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:892425>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "샷 텍스트에 명시되지 않은 인물들(양옆 배경에 엎드려 있는 사람들)이 등장함.",
     "fix_en": "Remove all the kneeling people in the background on the left and right.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "이전 샷(레퍼런스)에 있던 중앙의 평평한 제단 바위와 그 위의 뼈 목걸이가 사라졌으며, 중앙에 있던 단일 모닥불이 양옆 두 개로 변경되어 공간의 일관성이 깨짐.",
     "fix_en": "Restore the flat stone altar and bone necklace in the center, and replace the two side fires with a single central bonfire.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "지시문에 없는 낫 또는 갈고리 형태의 도구들이 왼쪽 바닥에 임의로 추가됨.",
     "fix_en": "Remove the curved sickle-like tools on the ground in the bottom left corner.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all the kneeling people in the background on the left and right.\n- Remove the curved sickle-like tools on the ground in the bottom left corner.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S105sh1"
  }
 },
 "S105sh7": {
  "input_fingerprint": "b7a31821b7c02ec4",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, full moon entering eclipse.\n\nSHOT TEXT (authoritative, Korean): 짙은 어둠이 하늘의 밝은 보름달 가장자리를 둥글게 가린 채 멈춰 있는 불길한 개기월식의 기괴한 밤하늘 광경.\n\nLOCATION (lock): In the open night sky above the forest ritual clearing and its fires. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The goats remain bound at the bloodstained altar, Youngran’s white robe remains blood-spattered, and Jihu’s pendant remains there.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 5,
   "A": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "하늘로 시선을 올려 인물을 자연스럽게 프레임에서 배제했으며, 보름달 가장자리가 가려지는 월식의 모습을 지시문대로 잘 묘사했습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "달의 묘사는 적절하나 프레임 하단에 제단 바위가 잡혔음에도 이전 샷에서 높이 들려 있던 제물과 인물의 팔이 사라져 연속성 오류가 발생했습니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "달이 붉은 고리 형태로 묘사와 다르고, 제단 전체가 프레임에 들어왔음에도 인물과 제물이 모두 사라지는 심각한 연속성 위반이 있습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S105sh3_sel.png"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "prev만 (배경 전용·공유 계획)",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S105sh3"
  },
  "lane_policy": "share_plan_prev_bgonly"
 },
 "S106sh2::bgfirst_bg": {
  "input_fingerprint": "a2f91c024cf3a39c",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 피칠갑을 한 고산들이 손에 쥔 하얀 사슴뿔 창을 바닥을 향해 늘어뜨린 채 서 있는 전신.\n\nLOCATION (lock): Immediately outside the island restaurant at night, at the top of the village lane.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Outside the open doorway at chest height, the forward dolly settles at a full-body distance directly before 고산들 as he completes a step out of the restaurant. His blood-covered figure occupies the central vertical band, while the lowered white antler spear runs beside his leg into the lower third and the open doorway remains visible behind him for the coming lateral reveal.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: restaurant doorway (Open) — The exterior side faces the camera, with the open passage revealing part of the hall behind 고산들; used as Frames 고산들's emergence and leaves a visible path into the hall for the next camera move; white antler spear (Carried in one lowered hand) — Its branching upper section is held at 고산들's side while the pointed end hangs toward the ground; used as Creates a pale downward line beside the blood-covered figure; restaurant hall floor (Partially visible) — Seen obliquely through the doorway behind 고산들; used as Provides restrained interior context through the open doorway.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 피칠갑을 한 고산들이 손에 쥔 하얀 사슴뿔 창을 바닥을 향해 늘어뜨린 채 서 있는 전신.\n\nLOCATION (lock): Immediately outside the island restaurant at night, at the top of the village lane.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Outside the open doorway at chest height, the forward dolly settles at a full-body distance directly before 고산들 as he completes a step out of the restaurant. His blood-covered figure occupies the central vertical band, while the lowered white antler spear runs beside his leg into the lower third and the open doorway remains visible behind him for the coming lateral reveal.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: restaurant doorway (Open) — The exterior side faces the camera, with the open passage revealing part of the hall behind 고산들; used as Frames 고산들's emergence and leaves a visible path into the hall for the next camera move; white antler spear (Carried in one lowered hand) — Its branching upper section is held at 고산들's side while the pointed end hangs toward the ground; used as Creates a pale downward line beside the blood-covered figure; restaurant hall floor (Partially visible) — Seen obliquely through the doorway behind 고산들; used as Provides restrained interior context through the open doorway.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S106sh2__bgfirst_bg.png",
  "asset_id": "babdd5a9-4504-4efc-a14e-0de54241c732",
  "input_asset_ids": [
   "3dea0135-b566-47d2-b24c-71469c2e3be0",
   "b756929e-cc3c-4e67-9df3-0cabc92e4864"
  ]
 },
 "S106sh2": {
  "input_fingerprint": "82e12e828645ba79",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 피칠갑을 한 고산들이 손에 쥔 하얀 사슴뿔 창을 바닥을 향해 늘어뜨린 채 서 있는 전신.\n\nLOCATION (lock): Immediately outside the island restaurant at night, at the top of the village lane. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): His battered corpse is sprawled motionless on the restaurant hall floor in a blood-soaked heap, with his torso and slack limbs collapsed against the boards.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Hwang’s mutilated corpse is sprawled motionless on the restaurant hall floor in a blood-soaked heap, with his torso and slack limbs collapsed against the boards. Sandeul is blood-covered, still bears his wounds, and carries the long white-antler-tipped spear taken from the shed shrine.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 피칠갑을 한 고산들이 손에 쥔 하얀 사슴뿔 창을 바닥을 향해 늘어뜨린 채 서 있는 전신.\n\nLOCATION (lock): Immediately outside the island restaurant at night, at the top of the village lane. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): His battered corpse is sprawled motionless on the restaurant hall floor in a blood-soaked heap, with his torso and slack limbs collapsed against the boards.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Hwang’s mutilated corpse is sprawled motionless on the restaurant hall floor in a blood-soaked heap, with his torso and slack limbs collapsed against the boards. Sandeul is blood-covered, still bears his wounds, and carries the long white-antler-tipped spear taken from the shed shrine.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 피칠갑을 한 고산들이 손에 쥔 하얀 사슴뿔 창을 바닥을 향해 늘어뜨린 채 서 있는 전신.\n\nLOCATION (lock): Immediately outside the island restaurant at night, at the top of the village lane. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): His battered corpse is sprawled motionless on the restaurant hall floor in a blood-soaked heap, with his torso and slack limbs collapsed against the boards.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Hwang’s mutilated corpse is sprawled motionless on the restaurant hall floor in a blood-soaked heap, with his torso and slack limbs collapsed against the boards. Sandeul is blood-covered, still bears his wounds, and carries the long white-antler-tipped spear taken from the shed shrine.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (한국인 40대 중반 남성, 짧게 깎은 검은 머리, 각진 얼굴, 날카로운 눈매) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S106sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S106sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:876576>"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_rocky_shore_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:876576>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "명시된 피칠갑 상태와 바닥을 향한 사슴뿔 창의 디테일, 식당 내부의 시신까지 매우 정확하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "장소와 전신 샷은 일치하나, 핵심 조건인 캐릭터의 피칠갑 상태가 전혀 반영되지 않았습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "명시된 피칠갑 상태와 바닥을 향한 사슴뿔 창의 디테일, 식당 내부의 시신까지 매우 정확하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "장소와 전신 샷은 일치하나, 핵심 조건인 캐릭터의 피칠갑 상태가 전혀 반영되지 않았습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "피칠갑을 한 모습, 하향된 사슴뿔 창, 식당 내부의 시신 등 프롬프트와 상태 지시문을 모두 훌륭하게 구현했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "전신 샷과 장소는 맞으나 샷 텍스트의 핵심인 '피칠갑을 한' 상태가 완전히 누락되었고 내부의 시신도 보이지 않습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "피칠갑을 한 모습, 하향된 사슴뿔 창, 식당 내부의 시신 등 프롬프트와 상태 지시문을 모두 훌륭하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "전신 샷과 장소는 맞으나 샷 텍스트의 핵심인 '피칠갑을 한' 상태가 완전히 누락되었고 내부의 시신도 보이지 않습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 8
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 15,
   "B": 8
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "명시된 피칠갑 상태와 바닥을 향한 사슴뿔 창의 디테일, 식당 내부의 시신까지 매우 정확하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "장소와 전신 샷은 일치하나, 핵심 조건인 캐릭터의 피칠갑 상태가 전혀 반영되지 않았습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_island_rocky_shore_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:876576>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "주인공이 레퍼런스에 있는 목 위까지 올라오는 검은색 라운드 티셔츠 대신 깃이 있고 앞이 파인 셔츠를 입고 가슴을 드러내고 있습니다.",
     "fix_en": "Change the character's inner shirt to a plain black crew-neck t-shirt that fully covers his chest, matching the character reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the character's inner shirt to a plain black crew-neck t-shirt that fully covers his chest, matching the character reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S106sh2__bgfirst_bg.png",
   "bg_asset_id": "babdd5a9-4504-4efc-a14e-0de54241c732",
   "bg_record_key": "S106sh2::bgfirst_bg",
   "chain_winner": true,
   "authority": "seed_bg"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S107sh3::bgfirst_bg": {
  "input_fingerprint": "601bb2af03cf7cd8",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 극심한 공포에 질려 눈물이 맺힌 채, 머리카락이 뒤로 휘날리며 숲길을 앞으로 내달리는 도중의 구창모 얼굴 클로즈업.\n\nLOCATION (lock): On a pitch-dark forest trail crowded by branches and undergrowth.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track backward at face height directly ahead of 구창모, tightening into a close frame while preserving the vertical jolt of his sprint and the backward stream of his hair. His tear-filled face occupies the center-left, with fragments of the dark path, approaching branches, and brush slipping past the frame edges to sustain forward momentum.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 구창모 in the middle-center of the frame, midground, moves toward forest path ahead.\n- KEY BACKGROUND ELEMENTS: 숲길 (깜깜해 방향을 분간하기 어려운 상태); used as Narrow moving backdrop behind the runner, retaining spatial evidence of his route; 나무 잔가지와 수풀 (구창모의 눈앞을 덮쳐오는 상태) — Their projecting ends angle into 구창모's running line and pass close to either side of his face; used as Branches and brush graze the frame edges as moving foreground obstructions.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 극심한 공포에 질려 눈물이 맺힌 채, 머리카락이 뒤로 휘날리며 숲길을 앞으로 내달리는 도중의 구창모 얼굴 클로즈업.\n\nLOCATION (lock): On a pitch-dark forest trail crowded by branches and undergrowth.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track backward at face height directly ahead of 구창모, tightening into a close frame while preserving the vertical jolt of his sprint and the backward stream of his hair. His tear-filled face occupies the center-left, with fragments of the dark path, approaching branches, and brush slipping past the frame edges to sustain forward momentum.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 구창모 in the middle-center of the frame, midground, moves toward forest path ahead.\n- KEY BACKGROUND ELEMENTS: 숲길 (깜깜해 방향을 분간하기 어려운 상태); used as Narrow moving backdrop behind the runner, retaining spatial evidence of his route; 나무 잔가지와 수풀 (구창모의 눈앞을 덮쳐오는 상태) — Their projecting ends angle into 구창모's running line and pass close to either side of his face; used as Branches and brush graze the frame edges as moving foreground obstructions.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S107sh3__bgfirst_bg.png",
  "asset_id": "e6c565f5-d6ce-4841-91fd-4be6a617c3e2",
  "input_asset_ids": [
   "2fbd3fac-1d01-4888-901c-1c7e4fbb4af8",
   "d598d2d4-ff94-4548-9220-fb3280d66002"
  ]
 },
 "S107sh3": {
  "input_fingerprint": "f622279c595eef16",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 극심한 공포에 질려 눈물이 맺힌 채, 머리카락이 뒤로 휘날리며 숲길을 앞으로 내달리는 도중의 구창모 얼굴 클로즈업.\n\nLOCATION (lock): On a pitch-dark forest trail crowded by branches and undergrowth. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (한국인 40대 중후반 남성, 호리한 얼굴, 짧은 검은 머리, 가무잡잡한 피부) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 극심한 공포에 질려 눈물이 맺힌 채, 머리카락이 뒤로 휘날리며 숲길을 앞으로 내달리는 도중의 구창모 얼굴 클로즈업.\n\nLOCATION (lock): On a pitch-dark forest trail crowded by branches and undergrowth. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (한국인 40대 중후반 남성, 호리한 얼굴, 짧은 검은 머리, 가무잡잡한 피부) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 극심한 공포에 질려 눈물이 맺힌 채, 머리카락이 뒤로 휘날리며 숲길을 앞으로 내달리는 도중의 구창모 얼굴 클로즈업.\n\nLOCATION (lock): On a pitch-dark forest trail crowded by branches and undergrowth. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (한국인 40대 중후반 남성, 호리한 얼굴, 짧은 검은 머리, 가무잡잡한 피부) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S107sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S107sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:730133>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L53B03.png"
    },
    {
     "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:730133>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "요구된 얼굴 클로즈업 구도 내에서 달리는 역동적인 포즈, 뒤로 휘날리는 머리카락, 극심한 공포와 눈물 등 텍스트의 모든 세부 사항을 훌륭하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "클로즈업 구도는 충족했으나, 달리는 동작, 뒤로 휘날리는 머리카락, 극심한 공포라는 핵심 액션과 감정 지시를 전혀 반영하지 못한 정적인 샷입니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "요구된 얼굴 클로즈업 구도 내에서 달리는 역동적인 포즈, 뒤로 휘날리는 머리카락, 극심한 공포와 눈물 등 텍스트의 모든 세부 사항을 훌륭하게 구현했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "클로즈업 구도는 충족했으나, 달리는 동작, 뒤로 휘날리는 머리카락, 극심한 공포라는 핵심 액션과 감정 지시를 전혀 반영하지 못한 정적인 샷입니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "공포에 질려 눈물을 흘리며 숲길을 달리는 역동적인 순간과 뒤로 휘날리는 머리카락 등 샷 텍스트의 구체적인 액션 요구사항을 매우 잘 구현했습니다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "얼굴 클로즈업과 눈물이 맺힌 표정은 좋으나, 앞으로 달려나가며 머리카락이 휘날리는 핵심적인 역동성이 전혀 묘사되지 않고 정지된 모습입니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "공포에 질려 눈물을 흘리며 숲길을 달리는 역동적인 순간과 뒤로 휘날리는 머리카락 등 샷 텍스트의 구체적인 액션 요구사항을 매우 잘 구현했습니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "얼굴 클로즈업과 눈물이 맺힌 표정은 좋으나, 앞으로 달려나가며 머리카락이 휘날리는 핵심적인 역동성이 전혀 묘사되지 않고 정지된 모습입니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 9
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 15,
   "B": 9
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "요구된 얼굴 클로즈업 구도 내에서 달리는 역동적인 포즈, 뒤로 휘날리는 머리카락, 극심한 공포와 눈물 등 텍스트의 모든 세부 사항을 훌륭하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "클로즈업 구도는 충족했으나, 달리는 동작, 뒤로 휘날리는 머리카락, 극심한 공포라는 핵심 액션과 감정 지시를 전혀 반영하지 못한 정적인 샷입니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L53B03.png"
   },
   {
    "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:730133>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "화면 왼쪽 하단에 보이는 주먹 쥔 손의 손가락 형태가 구분되지 않고 한 덩어리로 뭉개져 있습니다.",
     "fix_en": "Render distinct, anatomically correct fingers and knuckles on the clenched fist on the lower left.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "양 볼에 흐르는 눈물/땀 방울이 자연스러운 액체가 아니라 투명하고 두꺼운 젤리나 피부 돌기처럼 부자연스럽게 굳어 있습니다.",
     "fix_en": "Replace the thick, solid-looking clear blobs on the character's cheeks with natural, flat liquid tear or sweat streaks.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Render distinct, anatomically correct fingers and knuckles on the clenched fist on the lower left.\n- Replace the thick, solid-looking clear blobs on the character's cheeks with natural, flat liquid tear or sweat streaks.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S107sh3__bgfirst_bg.png",
   "bg_asset_id": "e6c565f5-d6ce-4841-91fd-4be6a617c3e2",
   "bg_record_key": "S107sh3::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S108sh6::bgfirst_bg": {
  "input_fingerprint": "986b1147bba64c49",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 천장에서 떨어져 내린 끈적한 검붉은 핏방울이 구창모의 손등에 막 닿아 맺혀 있는 정지된 찰나의 클로즈업.\n\nLOCATION (lock): Inside the cramped storage compartment of the slaughterhouse, beneath its dark ceiling.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at hand height in an isolated insert, aimed slightly upward beneath the drop so the exact point of contact sits near the center of the frame. The back of 구창모's trembling hand and part of his grip on the side-door handle provide scale, while the dark ceiling remains as a narrow upper background that motivates the coming upward tilt.\n- FRAMING SCALE: insert close-up on a detail\n- KEY BACKGROUND ELEMENTS: 어두운 천장 (위쪽이 어두운 상태); used as Upper-frame source direction for the falling liquid and the next tilt; 창고 쪽문 손잡이 (구창모가 꼭 붙들고 있는 상태) — The graspable side faces the camera at an oblique angle beneath 구창모's fingers; used as Scale reference beside the isolated hand and evidence of 구창모's attempted concealment; 검붉은 핏방울 (끈적한 방울이 손등에 막 닿아 맺히는 상태); used as Central focal detail at the instant of contact.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 천장에서 떨어져 내린 끈적한 검붉은 핏방울이 구창모의 손등에 막 닿아 맺혀 있는 정지된 찰나의 클로즈업.\n\nLOCATION (lock): Inside the cramped storage compartment of the slaughterhouse, beneath its dark ceiling.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at hand height in an isolated insert, aimed slightly upward beneath the drop so the exact point of contact sits near the center of the frame. The back of 구창모's trembling hand and part of his grip on the side-door handle provide scale, while the dark ceiling remains as a narrow upper background that motivates the coming upward tilt.\n- FRAMING SCALE: insert close-up on a detail\n- KEY BACKGROUND ELEMENTS: 어두운 천장 (위쪽이 어두운 상태); used as Upper-frame source direction for the falling liquid and the next tilt; 창고 쪽문 손잡이 (구창모가 꼭 붙들고 있는 상태) — The graspable side faces the camera at an oblique angle beneath 구창모's fingers; used as Scale reference beside the isolated hand and evidence of 구창모's attempted concealment; 검붉은 핏방울 (끈적한 방울이 손등에 막 닿아 맺히는 상태); used as Central focal detail at the instant of contact.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S108sh6__bgfirst_bg.png",
  "asset_id": "be867304-069e-4fd6-91ff-be4e2d03383c",
  "input_asset_ids": [
   "b151ea00-26a0-4d00-96a8-68e5a1f733c9",
   "dda47e7c-bf17-4f91-8e97-7bf6000f938b"
  ]
 },
 "S108sh6": {
  "input_fingerprint": "17f4f6e24bca67b0",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 천장에서 떨어져 내린 끈적한 검붉은 핏방울이 구창모의 손등에 막 닿아 맺혀 있는 정지된 찰나의 클로즈업.\n\nLOCATION (lock): Inside the cramped storage compartment of the slaughterhouse, beneath its dark ceiling. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Hwang’s corpse remains in the restaurant; Sandeul remains blood-covered and carries the antler-tipped spear while pursuing Changmo.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 천장에서 떨어져 내린 끈적한 검붉은 핏방울이 구창모의 손등에 막 닿아 맺혀 있는 정지된 찰나의 클로즈업.\n\nLOCATION (lock): Inside the cramped storage compartment of the slaughterhouse, beneath its dark ceiling. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Hwang’s corpse remains in the restaurant; Sandeul remains blood-covered and carries the antler-tipped spear while pursuing Changmo.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 천장에서 떨어져 내린 끈적한 검붉은 핏방울이 구창모의 손등에 막 닿아 맺혀 있는 정지된 찰나의 클로즈업.\n\nLOCATION (lock): Inside the cramped storage compartment of the slaughterhouse, beneath its dark ceiling. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Hwang’s corpse remains in the restaurant; Sandeul remains blood-covered and carries the antler-tipped spear while pursuing Changmo.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S108sh6__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S108sh6.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_ritual_clearing_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "피가 손등에 막 닿아 맺히는 찰나의 클로즈업을 정확히 연출했으며, 어두운 도축장의 분위기를 잘 살렸습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "떨어지는 핏방울이 손등에 닿는 순간을 정확히 포착하지 못했으며, 배경 요소가 텍스트의 장소 묘사와 부합하지 않습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "피가 손등에 막 닿아 맺히는 찰나의 클로즈업을 정확히 연출했으며, 어두운 도축장의 분위기를 잘 살렸습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "떨어지는 핏방울이 손등에 닿는 순간을 정확히 포착하지 못했으며, 배경 요소가 텍스트의 장소 묘사와 부합하지 않습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "손등에 핏방울이 맺히는 클로즈업 샷의 연출은 텍스트에 부합하나, 확정된 야외 숲/제단 레퍼런스를 무시하고 실내 도축장으로 배경을 임의 설정한 점이 감점 요인입니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "프롬프트에 없는 손잡이를 잡는 동작이 추가되었으며, A와 마찬가지로 기준이 되는 야외 배경 레퍼런스를 전혀 반영하지 못했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "손등에 핏방울이 맺히는 클로즈업 샷의 연출은 텍스트에 부합하나, 확정된 야외 숲/제단 레퍼런스를 무시하고 실내 도축장으로 배경을 임의 설정한 점이 감점 요인입니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "프롬프트에 없는 손잡이를 잡는 동작이 추가되었으며, A와 마찬가지로 기준이 되는 야외 배경 레퍼런스를 전혀 반영하지 못했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 8,
     "B": 12
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 8,
   "B": 12
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "피가 손등에 막 닿아 맺히는 찰나의 클로즈업을 정확히 연출했으며, 어두운 도축장의 분위기를 잘 살렸습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "떨어지는 핏방울이 손등에 닿는 순간을 정확히 포착하지 못했으며, 배경 요소가 텍스트의 장소 묘사와 부합하지 않습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_ritual_clearing_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "유일한 장소 기준인 레퍼런스 이미지(야외 숲 속의 돌 제단)를 무시하고 배경을 실내 도축장으로 묘사했습니다.",
     "fix_en": "Replace the entire indoor slaughterhouse background with the outdoor pine forest and stone structure from the reference photograph.",
     "needs_regeneration": true,
     "unfixable": true
    },
    {
     "issue_ko": "핏방울이 '막 닿아 맺혀 있는 정지된 찰나'라는 지문과 달리, 길게 이어지는 핏줄기가 떨어지고 이미 손등에 피가 넓게 고여 있습니다.",
     "fix_en": "Remove the continuous vertical stream of blood and the large pool on the hand, replacing them with a single, distinct blood drop just making contact with the skin.",
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the continuous vertical stream of blood and the large pool on the hand, replacing them with a single, distinct blood drop just making contact with the skin.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S108sh6__bgfirst_bg.png",
   "bg_asset_id": "be867304-069e-4fd6-91ff-be4e2d03383c",
   "bg_record_key": "S108sh6::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S108sh9": {
  "input_fingerprint": "79a57a37a0925d18",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 천장에 기괴하게 거꾸로 매달린 채 짐승처럼 살기 어린 눈빛으로 내려다보는 고산들의 섬뜩한 얼굴 클로즈업.\n\nLOCATION (lock): At the ceiling of the slaughterhouse storage compartment above the hidden man. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the cramped dark storeroom, blood drip, rough ceiling, sacks, and minimal night light. Exclude the hand-only blood-drop framing and reveal the man hanging above the officer.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandeul remains blood-covered and wounded, hanging from the ceiling above Changmo; the antler-tipped spear remains in his possession.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 7,
   "A": 5,
   "B": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "요구된 '얼굴 클로즈업' 샷 크기를 정확히 구현했으며, 지시대로 이전 샷의 피 묻은 손을 프레임에서 배제하여 프롬프트 충실도가 가장 높습니다."
   },
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "요구된 얼굴 클로즈업이 아닌 상반신이 모두 보이는 미디엄 샷으로 렌더링되어 최우선 지시사항인 샷 크기를 위반했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "얼굴 클로즈업 지시를 무시하고 샷을 넓게 잡았으며, 배제하라고 명시된 이전 샷의 피 묻은 손을 전경에 그대로 포함해 지시를 위반했습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S108sh6_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "캐릭터 얼굴 왼쪽 공간에 창이 손에 쥐어지지 않은 채 허공에 떠 있습니다.",
     "fix_en": "Remove the floating antler-tipped spear from the left side of the frame.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the floating antler-tipped spear from the left side of the frame.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S108sh6"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S109sh2": {
  "input_fingerprint": "b26d064376778b03",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 기괴하게 입을 벌린 채 몸을 한껏 뒤튼 검은 염소 떼의 뿔과 황금빛 눈동자 클로즈업.\n\nLOCATION (lock): Inside the fenced outdoor goat pens beside the forest slaughterhouse at night. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Changmo remains inside the slaughterhouse under Sandeul’s attack; the slaughterhouse bloodstains and tools remain in place.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person or any body part appears in frame, unless the shot text itself explicitly says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 3,
   "B": 8,
   "C": 6
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "염소의 뿔이 하나로 겹쳐지거나 여러 개로 왜곡되는 해부학적 오류가 있으며, 기괴하게 입을 벌리거나 몸을 뒤튼 묘사가 부족합니다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "입을 기괴하게 벌린 채 몸(목)을 한껏 뒤튼 검은 염소의 황금빛 눈동자를 클로즈업으로 가장 정확하고 생생하게 연출했습니다."
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "입을 벌린 염소와 황금빛 눈동자를 클로즈업으로 잘 담아냈으나, 몸을 한껏 뒤튼 역동적인 자세와 기괴함이 B에 비해 다소 부족합니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L53B02.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "중앙 염소의 동공이 염소 특유의 가로형 직사각형이 아닌 둥근 모양이며, 눈동자 색이 지문에 명시된 황금빛이 아닙니다.",
     "fix_en": "Change the central goat's round pupil into a horizontal, rectangular slit and recolor the iris to a vivid golden yellow."
    },
    {
     "issue_ko": "'몸을 한껏 뒤튼'이라는 지문 묘사와 달리, 중앙 염소의 목과 몸통 자세가 평범하고 충분히 비틀려 있지 않습니다.",
     "fix_en": "Warp the neck and upper shoulders of the central goat to make its posture appear severely contorted and twisted."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the central goat's round pupil into a horizontal, rectangular slit and recolor the iris to a vivid golden yellow.\n- Warp the neck and upper shoulders of the central goat to make its posture appear severely contorted and twisted.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "groupbg::의식바위_공터": {
  "input_fingerprint": "e718f63ac18b5240",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "의식바위_공터",
    "tags": [
     "S102sh17",
     "S105sh1",
     "S105sh3",
     "S105sh7",
     "S110sh21",
     "S110sh25",
     "S110sh41",
     "S111sh11",
     "S111sh14",
     "S111sh19",
     "S52sh2",
     "S52sh4",
     "S68sh1",
     "S68sh4",
     "S68sh7",
     "S73sh4",
     "S73sh5",
     "S78sh1",
     "S78sh4",
     "S78sh9"
    ]
   },
   "context_sig": "09ba6968e5ccd24e"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On the central sacrifice platform before the giant boulder in the moonlit ritual clearing.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 숲속 동굴 내부: 빛이 들지 않는 깊고 축축한 자연 동굴. 바닥에 인골과 토막 난 시체가 널려 있다. (특징: 울퉁불퉁하고 어두운 암석 벽과 천장; 천장에 빽빽이 매달린 수백 마리 박쥐 떼; 바닥에 흩어진 사람의 뼈와 토막 난 신체 부위; 바닥에 고인 피 웅덩이; 벽에 고정된 무거운 쇠사슬) / 금월도 숲속 만월제 도당: 숲 한가운데 둥글게 트인 공터. 거대한 바위와 제단이 있고 커다란 모닥불들이 타오른다. (특징: 성인 키 서너 배 높이의 거대한 암석 바위; 피 손자국으로 범벅이 된 바위 표면; 바위 앞 제단; 제단을 둥글게 감싸며 타오르는 커다란 모닥불들; 하늘 위 거대한 핏빛 보름달) / 금월도 해송 숲길 및 숲속: 해송이 울창하게 우거져 빛이 잘 들지 않는 빽빽하고 좁은 흙길과 숲의 심층부. (특징: 빽빽한 해송 나무 기둥들; 무성하고 어두운 수풀; 바닥에 깔린 흙과 솔잎; ‘수리영’이 적힌 이끼 낀 비석들; 비석 앞에 놓인 썩은 핏빛 육고기 덩어리) / 반월 식당 뒤뜰: 식당 건물 뒤편의 흙마당. 잘 꾸며진 텃밭과 평상이 있고 낡은 창고 건물이 보인다. (특징: 흙바닥 텃밭; 나무 평상; 말려둔 시래기 묶음; 허름한 창고 건물의 목재 외벽) / 숲속 도축장 내부: 벽에 도축용 기구들이 널려 있는 가건물 내부. 바닥과 벽면에 검붉은 핏자국이 선명하다. (특징: 도축용 칼과 뾰족한 금속 갈고리들; 피받이 양동이; 핏자국이 흥건한 도축대, 바닥, 벽면; 구석에 쌓인 피 묻은 포대자루 더미; 포대자루에서 쏟아진 토막 난 인체 부위) / 해안마을 선착장: 육지 쪽 항구의 선착장. 어선들이 정박해 있고 콘크리트 바닥에 폴리스 라인이 쳐져 있다. (특징: 콘크리트 항구 부두 바닥; 드럼통 모닥불; 시체 주변의 노란색 폴리스 라인; 구경꾼 무리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 숲속 만월제 도당 - 밤\n- 산 속 숲 한 가운데 공터처럼 뻥 뚫린 공간.\n- 실외. 숲속 만월제 도당 - 밤\n- 실외. 숲 속 만월제 도당 - 밤\n- 실외. 숲 속 만월제 도당 - 밤\n- 실외. 숲속 만월제 도당 - 밤\n- 실외. 숲 속 만월제 도당 - 밤\n- 같은 장소 - 잠시 후\n\nTIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On the central sacrifice platform before the giant boulder in the moonlit ritual clearing.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 숲속 동굴 내부: 빛이 들지 않는 깊고 축축한 자연 동굴. 바닥에 인골과 토막 난 시체가 널려 있다. (특징: 울퉁불퉁하고 어두운 암석 벽과 천장; 천장에 빽빽이 매달린 수백 마리 박쥐 떼; 바닥에 흩어진 사람의 뼈와 토막 난 신체 부위; 바닥에 고인 피 웅덩이; 벽에 고정된 무거운 쇠사슬) / 금월도 숲속 만월제 도당: 숲 한가운데 둥글게 트인 공터. 거대한 바위와 제단이 있고 커다란 모닥불들이 타오른다. (특징: 성인 키 서너 배 높이의 거대한 암석 바위; 피 손자국으로 범벅이 된 바위 표면; 바위 앞 제단; 제단을 둥글게 감싸며 타오르는 커다란 모닥불들; 하늘 위 거대한 핏빛 보름달) / 금월도 해송 숲길 및 숲속: 해송이 울창하게 우거져 빛이 잘 들지 않는 빽빽하고 좁은 흙길과 숲의 심층부. (특징: 빽빽한 해송 나무 기둥들; 무성하고 어두운 수풀; 바닥에 깔린 흙과 솔잎; ‘수리영’이 적힌 이끼 낀 비석들; 비석 앞에 놓인 썩은 핏빛 육고기 덩어리) / 반월 식당 뒤뜰: 식당 건물 뒤편의 흙마당. 잘 꾸며진 텃밭과 평상이 있고 낡은 창고 건물이 보인다. (특징: 흙바닥 텃밭; 나무 평상; 말려둔 시래기 묶음; 허름한 창고 건물의 목재 외벽) / 숲속 도축장 내부: 벽에 도축용 기구들이 널려 있는 가건물 내부. 바닥과 벽면에 검붉은 핏자국이 선명하다. (특징: 도축용 칼과 뾰족한 금속 갈고리들; 피받이 양동이; 핏자국이 흥건한 도축대, 바닥, 벽면; 구석에 쌓인 피 묻은 포대자루 더미; 포대자루에서 쏟아진 토막 난 인체 부위) / 해안마을 선착장: 육지 쪽 항구의 선착장. 어선들이 정박해 있고 콘크리트 바닥에 폴리스 라인이 쳐져 있다. (특징: 콘크리트 항구 부두 바닥; 드럼통 모닥불; 시체 주변의 노란색 폴리스 라인; 구경꾼 무리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 숲속 만월제 도당 - 밤\n- 산 속 숲 한 가운데 공터처럼 뻥 뚫린 공간.\n- 실외. 숲속 만월제 도당 - 밤\n- 실외. 숲 속 만월제 도당 - 밤\n- 실외. 숲 속 만월제 도당 - 밤\n- 실외. 숲속 만월제 도당 - 밤\n- 실외. 숲 속 만월제 도당 - 밤\n- 같은 장소 - 잠시 후\n\nTIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_의식바위_공터_09f08a.png",
  "asset_id": "25beca28-ae82-41ac-abd3-dcfe81125188",
  "input_asset_ids": [
   "a0c76d83-4137-466d-8d4d-80ca8ee646a8"
  ],
  "origin_tag": "S110sh21",
  "place_text": "On the central sacrifice platform before the giant boulder in the moonlit ritual clearing.",
  "origin_inputs": {
   "place_text": "On the central sacrifice platform before the giant boulder in the moonlit ritual clearing.",
   "time_of_day_en": "night, total lunar eclipse and blood-red moon",
   "conti_asset_id": "a0c76d83-4137-466d-8d4d-80ca8ee646a8"
  }
 },
 "S110sh21::bgfirst_bg": {
  "input_fingerprint": "08c8a8e3655181e3",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 얼굴에 문자가 새겨진 김지후(강수리영)가 섬뜩한 황금빛 눈동자를 번뜩이며 두 눈을 크게 부릅뜨고 있는 얼굴 클로즈업.\n\nLOCATION (lock): On the central sacrifice platform before the giant boulder in the moonlit ritual clearing.\n\nTIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: End the frontal dolly-in at an extreme close distance at 김지후(강수리영)'s face height, using the direct axis as an occult confrontation rather than a neutral portrait. Her widened golden eyes sit across the upper center while the spreading characters remain legible over her cheeks and forehead, with only a thin trace of the ritual space at the frame boundary.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 김지후(강수리영) in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 얼굴에 퍼진 문자 문양 (얼굴 전체를 뒤덮기 시작한 상태) — The marked skin faces the camera directly, showing black ink-like Hebrew characters spreading across the forehead, cheeks, and facial contours; used as Visible transformation detail distributed across the subject's face; 중앙 제단과 거대한 바위 (김지후(강수리영)가 제단에 포박된 상태) — The altar-facing side and the rock behind it fall outside sharp focus at the frame margin; used as Minimal out-of-focus contextual edge behind the bound subject.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 얼굴에 문자가 새겨진 김지후(강수리영)가 섬뜩한 황금빛 눈동자를 번뜩이며 두 눈을 크게 부릅뜨고 있는 얼굴 클로즈업.\n\nLOCATION (lock): On the central sacrifice platform before the giant boulder in the moonlit ritual clearing.\n\nTIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: End the frontal dolly-in at an extreme close distance at 김지후(강수리영)'s face height, using the direct axis as an occult confrontation rather than a neutral portrait. Her widened golden eyes sit across the upper center while the spreading characters remain legible over her cheeks and forehead, with only a thin trace of the ritual space at the frame boundary.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 김지후(강수리영) in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 얼굴에 퍼진 문자 문양 (얼굴 전체를 뒤덮기 시작한 상태) — The marked skin faces the camera directly, showing black ink-like Hebrew characters spreading across the forehead, cheeks, and facial contours; used as Visible transformation detail distributed across the subject's face; 중앙 제단과 거대한 바위 (김지후(강수리영)가 제단에 포박된 상태) — The altar-facing side and the rock behind it fall outside sharp focus at the frame margin; used as Minimal out-of-focus contextual edge behind the bound subject.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S110sh21__bgfirst_bg.png",
  "asset_id": "f19fe2da-9903-416c-8ed8-1162ca4d6c37",
  "input_asset_ids": [
   "a0c76d83-4137-466d-8d4d-80ca8ee646a8",
   "25beca28-ae82-41ac-abd3-dcfe81125188"
  ]
 },
 "S110sh21": {
  "input_fingerprint": "b86bf7b79af87350",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nSHOT TEXT (authoritative, Korean): 얼굴에 문자가 새겨진 김지후(강수리영)가 섬뜩한 황금빛 눈동자를 번뜩이며 두 눈을 크게 부릅뜨고 있는 얼굴 클로즈업.\n\nLOCATION (lock): On the central sacrifice platform before the giant boulder in the moonlit ritual clearing. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is stretched out on the central altar in front of the rock with both wrists and both ankles immobilized by thick leather straps, her body aligned along the altar.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu is stretched out on the central altar in front of the rock with both wrists and both ankles immobilized by thick leather straps, her body aligned along the altar. Her body and face are covered in black script-like markings and her eyes are transformed to golden rectangular pupils. Her antler-bone pendant remains at the altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nSHOT TEXT (authoritative, Korean): 얼굴에 문자가 새겨진 김지후(강수리영)가 섬뜩한 황금빛 눈동자를 번뜩이며 두 눈을 크게 부릅뜨고 있는 얼굴 클로즈업.\n\nLOCATION (lock): On the central sacrifice platform before the giant boulder in the moonlit ritual clearing. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is stretched out on the central altar in front of the rock with both wrists and both ankles immobilized by thick leather straps, her body aligned along the altar.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu is stretched out on the central altar in front of the rock with both wrists and both ankles immobilized by thick leather straps, her body aligned along the altar. Her body and face are covered in black script-like markings and her eyes are transformed to golden rectangular pupils. Her antler-bone pendant remains at the altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nSHOT TEXT (authoritative, Korean): 얼굴에 문자가 새겨진 김지후(강수리영)가 섬뜩한 황금빛 눈동자를 번뜩이며 두 눈을 크게 부릅뜨고 있는 얼굴 클로즈업.\n\nLOCATION (lock): On the central sacrifice platform before the giant boulder in the moonlit ritual clearing. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): She is stretched out on the central altar in front of the rock with both wrists and both ankles immobilized by thick leather straps, her body aligned along the altar.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu is stretched out on the central altar in front of the rock with both wrists and both ankles immobilized by thick leather straps, her body aligned along the altar. Her body and face are covered in black script-like markings and her eyes are transformed to golden rectangular pupils. Her antler-bone pendant remains at the altar.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S110sh21__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S110sh21.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_의식바위_공터_09f08a.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "프롬프트에 명시되지 않은 인물들이 배경에 등장하여 치명적인 오류(Hard Violation)를 범했으며, 묶여 있어야 할 손목의 위치와 자세가 부자연스럽습니다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "얼굴의 문자, 황금빛 직사각형 눈동자, 제단 위에 결박된 자세와 뼈 펜던트 등 명시된 조건과 카메라 구도를 매우 충실하게 구현했습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "프롬프트에 명시되지 않은 인물들이 배경에 등장하여 치명적인 오류(Hard Violation)를 범했으며, 묶여 있어야 할 손목의 위치와 자세가 부자연스럽습니다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "얼굴의 문자, 황금빛 직사각형 눈동자, 제단 위에 결박된 자세와 뼈 펜던트 등 명시된 조건과 카메라 구도를 매우 충실하게 구현했습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "얼굴 클로즈업, 황금빛 눈동자, 얼굴의 문자 등 샷 텍스트를 정확히 구현했으며, 요구된 제단 위 결박 상태와 사슴뼈 펜던트, 주변에 다른 인물이 없는 조건까지 모두 충실히 반영했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "프롬프트에 없는 주변 인물들을 배경에 추가한 치명적 위반(invented people)이 발생했으며, 결박된 상태임에도 손에 힘을 주고 있는 등 중력과 무력화 지침을 어겼습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "얼굴 클로즈업, 황금빛 눈동자, 얼굴의 문자 등 샷 텍스트를 정확히 구현했으며, 요구된 제단 위 결박 상태와 사슴뼈 펜던트, 주변에 다른 인물이 없는 조건까지 모두 충실히 반영했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "프롬프트에 없는 주변 인물들을 배경에 추가한 치명적 위반(invented people)이 발생했으며, 결박된 상태임에도 손에 힘을 주고 있는 등 중력과 무력화 지침을 어겼습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 6,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "프롬프트에 명시되지 않은 인물들이 배경에 등장하여 치명적인 오류(Hard Violation)를 범했으며, 묶여 있어야 할 손목의 위치와 자세가 부자연스럽습니다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "얼굴의 문자, 황금빛 직사각형 눈동자, 제단 위에 결박된 자세와 뼈 펜던트 등 명시된 조건과 카메라 구도를 매우 충실하게 구현했습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_의식바위_공터_09f08a.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "샷 텍스트는 '얼굴 클로즈업'을 요구했으나, 현재 이미지는 어깨와 양팔 전체가 포함된 넓은 구도입니다.",
     "fix_en": "The framing is too wide; crop the image tightly around the character's face to achieve the requested close-up.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "캐릭터의 동공이 설정에 명시된 직사각형이 아니라 세로로 찢어진 타원형(고양이 눈) 형태입니다.",
     "fix_en": "Modify the pupils so they are clearly defined golden rectangles.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Modify the pupils so they are clearly defined golden rectangles.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S110sh21__bgfirst_bg.png",
   "bg_asset_id": "f19fe2da-9903-416c-8ed8-1162ca4d6c37",
   "bg_record_key": "S110sh21::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "의식바위_공터",
   "groupbg_asset_id": "25beca28-ae82-41ac-abd3-dcfe81125188"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S110sh25": {
  "input_fingerprint": "af1f4fa666c74a26",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night, total lunar eclipse and blood moon.\n\nSHOT TEXT (authoritative, Korean): 돌아서던 강영란(흰 머리 여자)의 가슴팍을 고산들이 쥔 하얀 사슴뿔 창이 거칠게 푹 꿰뚫은 순간.\n\nLOCATION (lock): The sacrificial altar directly before the boulder amid the firelit ritual gathering. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height on the lateral line of 고산들’s charge, the tracking camera holds a tightened three-quarter profile at the instant the white antler spear pierces 강영란(흰 머리 여자). Her struck torso occupies the right-center, 고산들’s driving hands and shaft cross the middle, and his forward momentum remains visible at the left without obscuring their locked eye line.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 고산들 in the middle-left of the frame, midground, moves toward 강영란(흰 머리 여자); 강영란(흰 머리 여자) in the middle-right of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 하얀 사슴뿔 창 (Driven through 강영란’s chest) — The shaft points from frame left through 강영란’s chest toward the altar axis; used as The principal diagonal crossing between the attacker’s hands and the victim’s torso; 제단 (Located on the spear’s continuing path) — Its front and side edges recede behind 강영란; used as Partial ritual-space context behind the impact line.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The red moon’s transformed illumination renders the impact in severe crimson high contrast against the surrounding night.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same ritual clearing, bloodied rock, bonfires, red moonlight, worshippers, and ceremonial clothing. Exclude the earlier static sacrifice tableau; show the returning man driving the white antler spear through the elder toward the bound woman.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandel retains the bloody stab and gunshot wounds while wielding the spear; its white-antler tip is driven through Youngran’s bloodstained white robe and chest toward Jihu. Jihu remains marked and golden-eyed.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 고산들 right now, so 고산들's hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 고산들: its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (Korean woman in her late 60s, neatly twisted white hair, graceful dignified face); 고산들 (Korean man in his mid-40s, closely cropped dark hair, cold angular face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "initial_roll_all_fail": true,
  "readings": [
   {
    "built_space": "가슴 높이의 측면 카메라에 가까운 미디엄 숏이다. 고산들은 중간 왼쪽, 강영란은 중간 오른쪽에 있고, 돌 제단과 거대한 바위가 강영란 뒤의 창 진행축에 놓인다. 다만 제단 위에 이전 장면의 결박된 여성이 그대로 보이고 주변에 여러 의식 참가자가 추가되어 있다.",
    "direction": "고산들은 왼쪽에서 오른쪽으로 창을 밀며 강영란과 눈을 맞춘다. 양손 사이의 자루가 화면 중앙을 가로질러 강영란의 가슴을 실제로 관통하고, 끝은 오른쪽 뒤의 제단 방향으로 이어진다. 세 후보 중 창의 조준과 관통 순간은 유일하게 정확하다.",
    "entities": "고산들은 한국인 중년 남성으로 보이며 짧은 머리, 각진 얼굴, 녹색 조끼와 체크 셔츠가 인물 참고와 대체로 맞고 상처 분장도 있다. 강영란은 한국인 노년 여성이고 흰 머리 쪽과 피 묻은 흰 예복이 맞지만 얼굴과 소매 장식은 참고와 다소 다르다. 창은 목재 자루이나 관통 후 보이는 끝이 흰 사슴뿔의 형태·색보다 붉은 금속성 촉처럼 읽힌다. 명시된 두 사람 외에 제단 위 여성과 주변 참가자들이 보인다.",
    "hard_violations": [
     "장면에 허용된 강영란과 고산들 외에 제단 위 결박된 여성 1명과 주변 의식 참가자 여러 명이 추가되어 있다."
    ],
    "label": "A",
    "physics": "고산들은 두 손으로 자루를 확실히 움켜쥐고 앞으로 밀며 상체를 기울여 추진력을 만든다. 창은 강영란의 가슴을 지나 등 뒤로 나온 것으로 보이며 양쪽 의상에 접촉한다. 두 주연은 지면에 선 자세로 체중이 지지되고, 창도 손과 관통 접촉으로 지지되어 부유하는 물체는 없다."
   },
   {
    "built_space": "미디엄 숏 안에 돌 제단과 뒤쪽 바위, 양옆 모닥불이 보이며 공간 자체는 의식 장소로 읽힌다. 그러나 강영란이 왼쪽, 고산들이 오른쪽에 있어 지정된 화면 배치와 공격 진행축이 완전히 반대다. 제단은 두 사람 사이 뒤에 있지만 창의 연장선이 요구된 제단 축을 명확히 통과하지 않는다.",
    "direction": "고산들은 오른쪽에서 왼쪽으로 창을 겨누고 강영란과 눈을 맞춘다. 흰 갈래 촉은 강영란의 가슴 앞에서 멈춰 있어 표적에 닿지도, 가슴을 관통하지도 않는다. 따라서 authoritative shot의 정확한 순간이 아니다.",
    "entities": "두 사람 모두 한국인으로 보이고 강영란은 노년 여성, 고산들은 중년 남성으로 읽힌다. 하지만 고산들은 참고의 짧은 머리와 현대 작업복 대신 상투·검은 조선식 복장을 입었고 정확한 얼굴도 다르다. 강영란 역시 참고의 피 묻고 장식된 예복 대신 깨끗하고 단순한 흰옷이며 머리도 과도한 땋은 형태다. 창은 목재 자루와 흰 뿔 재질이지만 참고의 단일 뾰족 촉이 아니라 큰 이지창 형태다.",
    "hard_violations": [],
    "label": "B",
    "physics": "고산들이 두 손으로 창을 잡고 두 발로 선 자세여서 사람과 소품의 지지는 가능하다. 다만 창끝과 강영란 사이에 틈이 있고 강영란에게 충격 반응도 없어, 관통 동작을 만들어 낸 물리적 접촉이나 결과가 전혀 없다."
   },
   {
    "built_space": "고산들이 중간 왼쪽, 강영란이 중간 오른쪽에 있어 기본 좌우 배치는 가장 정확하다. 돌 제단의 전면·측면과 거대한 바위가 강영란 뒤에 놓이고 모닥불도 의식 공간을 형성한다. 다만 무릎 가까이까지 보이는 다소 넓은 미디엄 숏이며, 요구된 긴박한 가슴 높이의 타이트한 3/4 프로필보다 느슨하다.",
    "direction": "고산들은 왼쪽에서 오른쪽으로 창을 내밀며 강영란을 바라보고, 강영란도 그를 마주 본다. 창끝은 강영란의 가슴 방향을 향하지만 몸 앞에서 상당한 간격을 두고 멈춰 있어 표적에 닿지 않으며 관통도 없다.",
    "entities": "두 사람은 각각 한국인 중년 남성과 노년 여성으로 읽힌다. 그러나 고산들은 참고의 짧은 머리와 현대 녹색 조끼·체크 셔츠 대신 상투와 밝은 조선식 옷을 입었고, 상처 분장도 빠졌다. 강영란은 흰 머리 쪽과 흰옷은 맞지만 참고의 얼굴, 피 묻은 장식 예복과 상태가 충분히 유지되지 않았다. 창은 목재 자루와 흰색 단일 촉으로 B보다 참고 소품에 가깝다.",
    "hard_violations": [],
    "label": "C",
    "physics": "고산들은 넓게 벌린 두 다리로 지면을 딛고 양손으로 창을 지지하여 공격 준비 자세 자체는 가능하다. 강영란도 지면에 서 있어 부유 문제는 없다. 그러나 창끝이 몸과 접촉하지 않고 두 사람 모두 정지 대치 자세이므로, 거칠게 찔러 관통한 순간의 충격과 전진 운동은 성립하지 않는다."
   }
  ],
  "totals": {
   "A": 1,
   "B": 2,
   "C": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 1,
    "verdict_ko": "좌우 배치·시선·가슴 관통 순간은 가장 정확하지만, 허용되지 않은 제단 위 여성과 여러 참가자를 추가한 중대한 인물 중복 때문에 실격이다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "창이 가슴에 닿지 않은 데다 인물 좌우와 공격 방향이 반대이고, 두 사람의 의상 및 창 형태도 참고와 크게 다르다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "고산들 왼쪽·강영란 오른쪽과 제단 위치는 가장 잘 맞지만, 창이 가슴 앞에서 멈춰 핵심인 관통 순간을 구현하지 못했다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S110sh21_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1379088>"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903959>"
   },
   {
    "label": "PROP REFERENCE — 창(하얀 사슴뿔 창 포함): the exact object appearing in this shot; match its look, material and wear exactly.",
    "path": "<bytes:942357>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "창이 여자의 가슴을 꿰뚫지 않고 두 사람이 떨어져 서 있어 샷의 핵심 액션이 누락되었습니다.",
     "fix_en": "Lengthen the spear shaft rightward so the antler tip buries into the woman's chest with a bloodstain at the entry point, preserving the people present and their positions, their clothing, the set, the light, and the framing.",
     "needs_regeneration": true
    },
    {
     "issue_ko": "남자가 레퍼런스의 현대식 작업복(비니, 조끼, 셔츠) 대신 조선시대 한복과 상투 차림을 하고 있습니다.",
     "fix_en": "Replace the man's topknot and hanbok with a dark beanie, olive canvas vest, and blue flannel shirt, preserving the people present and their positions, the woman's clothing, the set, the light, and the framing."
    },
    {
     "issue_ko": "여자의 옷에 핏자국이 전혀 없으며, 레퍼런스의 족두리와 색동 소매가 누락되었습니다.",
     "fix_en": "Add heavy bloodstains to the woman's face and robe, change her sleeve to feature colorful horizontal stripes, and place a black-and-gold jokduri crown on her head, preserving the people present and their positions, the man's clothing, the set, the light, and the framing."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 5,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Lengthen the spear shaft rightward so the antler tip buries into the woman's chest with a bloodstain at the entry point, preserving the people present and their positions, their clothing, the set, the light, and the framing.\n- Replace the man's topknot and hanbok with a dark beanie, olive canvas vest, and blue flannel shirt, preserving the people present and their positions, the woman's clothing, the set, the light, and the framing.\n- Add heavy bloodstains to the woman's face and robe, change her sleeve to feature colorful horizontal stripes, and place a black-and-gold jokduri crown on her head, preserving the people present and their positions, the man's clothing, the set, the light, and the framing.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": true,
    "readings": [
     {
      "label": "A",
      "direction": "고산들로 보이는 남자는 화면 좌측 중경에서 창을 우측으로 겨누지만, 창끝(하얀 사슴뿔 촉)은 강영란의 가슴에 닿지도 않고 그녀의 몸 훨씬 앞(화면 중앙 빈 공간, 제단 쪽)에서 멈춰 있다. 관통은 전혀 일어나지 않았다. 두 사람의 시선은 서로를 향해 맞물려 있으나, 샷 텍스트가 요구한 '가슴팍을 푹 꿰뚫은 순간'은 화면에 없다. 창 축은 좌하→우상 대각이지만 목표에 착지하지 않는다.",
      "built_space": "제단(돌 단)은 중앙 배경에 있고 그 뒤로 피 묻은 거석, 좌우로 화톳불 화로와 제기 상이 배치 — 이전 스틸의 장소와 대체로 일치. 다만 창이 계속 가는 경로상 제단이 놓여야 한다는 지시는 성립하나, 강영란 뒤로 제단 앞·옆면이 물러나는 배치가 아니라 두 인물 사이 중앙에 제단이 놓여 충돌선이 비어 보인다. 달(붉은 달)은 프레임에 없고 붉은 기가 약하다.",
      "entities": "여성: 백발을 단정히 틀어 올린 60대 후반 한국 여성으로 캐릭터 레퍼런스와 얼굴·머리는 근접하나, 레퍼런스의 족두리(장식 모자), 색동 소매단, 노리개, 피 얼룩진 흰 도포가 모두 사라지고 밋밋한 순백 한복만 남았다 — 의상 락 위반. 얼굴의 상처·피멍도 없음. 남자: 고산들 레퍼런스(비니, 올리브 조끼, 체크 셔츠, 현대 작업복)가 아니라 상투에 망건을 쓴 조선시대 백색 저고리 차림 — 인물 동일성·의상이 완전히 다르다. 창은 나무 자루+흰 뿔촉으로 프롭 레퍼런스와 유사하나 촉이 다소 길고 금속처럼 매끈하다. 산들의 자상·총상 흔적 없음.",
      "hard_violations": [
       "샷 텍스트가 명시한 '가슴팍을 푹 꿰뚫은 순간'이 전혀 구현되지 않음(창끝이 피사체에 닿지 않음) — 지정된 결정적 액션 미이행",
       "고산들이 레퍼런스의 현대 인물이 아닌 조선시대 복장·상투 인물로 대체되어 지정 인물이 아닌 사람으로 읽힘"
      ],
      "physics": "남자는 두 발을 벌린 런지 자세로 지면에 접지, 두 손으로 자루를 쥐고 있어 지지 자체는 정상. 여자는 두 발로 서 있고 팔은 늘어뜨린 상태 — 관통 충격을 받은 몸의 반응(뒤로 밀림, 흉부 함몰, 손의 반응, 출혈)이 전무하다. 즉 물리적으로는 '찌르기 직전 대치' 장면이지 '관통 순간'이 아니다. 공중에 뜬 물체는 없음."
     },
     {
      "label": "B",
      "direction": "고산들은 좌측 중경에서 두 손으로 창을 쥐고 우측 강영란을 향해 돌진 자세를 취하며, 창끝(흰 사슴뿔 촉)은 강영란의 가슴 높이에 거의 닿아 있다 — 촉 끝이 그녀의 흰 도포 가슴팍 앞에 접해 있고 가슴에는 검붉은 얼룩이 번져 있다. 다만 실제로 몸을 '꿰뚫고' 나온 상태는 아니고 접촉 직전/막 닿은 순간에 가깝다. 두 인물의 시선은 서로 맞물려 있고 창 자루가 프레임 좌하→중앙 대각선을 이룬다.",
      "built_space": "이전 스틸의 피 묻은 거석, 돌 제단, 화톳불 화로(좌·우), 제기 상, 어두운 숲이 그대로 유지 — 장소 락 일치. 제단은 강영란 뒤/좌측으로 물러나며 창의 연장선상에 위치. 인물 대비 배경 요소 스케일 자연스럽고 과장된 전경화 없음. 붉은 달은 프레임 밖이나 붉은 화광이 지배.",
      "entities": "여성: 백발 쪽머리, 60대 후반 한국 여성, 레퍼런스의 족두리(꽃장식 모자), 색동 소매단, 노리개, 얼굴의 상처·먹 자국까지 재현되어 캐릭터 레퍼런스와 잘 일치. 흰 도포의 가슴에 피 얼룩 있음. 남자: 갈색 비니, 올리브 작업 조끼, 청/네이비 체크 셔츠, 작업 바지 — 고산들 레퍼런스 의상과 정확히 일치, 40대 중반 각진 얼굴. 창은 나무 자루에 끈으로 감긴 흰 뿔촉으로 프롭 레퍼런스와 부합. 산들의 자상·총상 유혈 흔적은 뚜렷하지 않음.",
      "hard_violations": [],
      "physics": "남자는 앞발을 앞으로 내딛고 뒷발로 밀어내는 런지로 지면에 확실히 접지, 양손이 자루를 정확히 파지 — 돌진의 하중 이동이 읽힌다. 여자는 두 발로 지면에 서 있고 몸이 반쯤 돌아선 상태로 창을 맞이한다. 공중에 뜬 무지지 물체나 인체는 없다. 다만 관통 충격에 대한 몸의 반작용(밀림·경직)은 약하다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "고산들을 조선시대 상투·백의 인물로 바꿔 캐릭터 레퍼런스를 통째로 잃었고, 강영란의 족두리·색동 소매·노리개·핏자국까지 삭제했으며, 창끝은 가슴에서 한참 떨어져 '꿰뚫은 순간'이 아예 없다 — 계약이 요구하지 않은 변경만 늘고 핵심 액션은 미이행."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "고산들의 비니·올리브 조끼·체크 셔츠와 강영란의 족두리·색동 소매·상처가 레퍼런스대로 유지되고 창끝이 가슴팍에 닿아 충돌선이 성립 — 관통 깊이와 피격 반작용이 부족하고 산들의 자상·총상이 안 보이는 점은 감점이나 두 후보 중 계약 이행도가 명백히 높다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "고산들을 조선시대 상투·백의 인물로 바꿔 캐릭터 레퍼런스를 통째로 잃었고, 강영란의 족두리·색동 소매·노리개·핏자국까지 삭제했으며, 창끝은 가슴에서 한참 떨어져 '꿰뚫은 순간'이 아예 없다 — 계약이 요구하지 않은 변경만 늘고 핵심 액션은 미이행."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "고산들의 비니·올리브 조끼·체크 셔츠와 강영란의 족두리·색동 소매·상처가 레퍼런스대로 유지되고 창끝이 가슴팍에 닿아 충돌선이 성립 — 관통 깊이와 피격 반작용이 부족하고 산들의 자상·총상이 안 보이는 점은 감점이나 두 후보 중 계약 이행도가 명백히 높다."
     }
    ],
    "all_candidates_fail": true,
    "readings": [
     {
      "label": "A",
      "direction": "고산들로 보이는 남자는 화면 좌측 중경에서 창을 우측으로 겨누지만, 창끝(하얀 사슴뿔 촉)은 강영란의 가슴에 닿지도 않고 그녀의 몸 훨씬 앞(화면 중앙 빈 공간, 제단 쪽)에서 멈춰 있다. 관통은 전혀 일어나지 않았다. 두 사람의 시선은 서로를 향해 맞물려 있으나, 샷 텍스트가 요구한 '가슴팍을 푹 꿰뚫은 순간'은 화면에 없다. 창 축은 좌하→우상 대각이지만 목표에 착지하지 않는다.",
      "built_space": "제단(돌 단)은 중앙 배경에 있고 그 뒤로 피 묻은 거석, 좌우로 화톳불 화로와 제기 상이 배치 — 이전 스틸의 장소와 대체로 일치. 다만 창이 계속 가는 경로상 제단이 놓여야 한다는 지시는 성립하나, 강영란 뒤로 제단 앞·옆면이 물러나는 배치가 아니라 두 인물 사이 중앙에 제단이 놓여 충돌선이 비어 보인다. 달(붉은 달)은 프레임에 없고 붉은 기가 약하다.",
      "entities": "여성: 백발을 단정히 틀어 올린 60대 후반 한국 여성으로 캐릭터 레퍼런스와 얼굴·머리는 근접하나, 레퍼런스의 족두리(장식 모자), 색동 소매단, 노리개, 피 얼룩진 흰 도포가 모두 사라지고 밋밋한 순백 한복만 남았다 — 의상 락 위반. 얼굴의 상처·피멍도 없음. 남자: 고산들 레퍼런스(비니, 올리브 조끼, 체크 셔츠, 현대 작업복)가 아니라 상투에 망건을 쓴 조선시대 백색 저고리 차림 — 인물 동일성·의상이 완전히 다르다. 창은 나무 자루+흰 뿔촉으로 프롭 레퍼런스와 유사하나 촉이 다소 길고 금속처럼 매끈하다. 산들의 자상·총상 흔적 없음.",
      "hard_violations": [
       "샷 텍스트가 명시한 '가슴팍을 푹 꿰뚫은 순간'이 전혀 구현되지 않음(창끝이 피사체에 닿지 않음) — 지정된 결정적 액션 미이행",
       "고산들이 레퍼런스의 현대 인물이 아닌 조선시대 복장·상투 인물로 대체되어 지정 인물이 아닌 사람으로 읽힘"
      ],
      "physics": "남자는 두 발을 벌린 런지 자세로 지면에 접지, 두 손으로 자루를 쥐고 있어 지지 자체는 정상. 여자는 두 발로 서 있고 팔은 늘어뜨린 상태 — 관통 충격을 받은 몸의 반응(뒤로 밀림, 흉부 함몰, 손의 반응, 출혈)이 전무하다. 즉 물리적으로는 '찌르기 직전 대치' 장면이지 '관통 순간'이 아니다. 공중에 뜬 물체는 없음."
     },
     {
      "label": "B",
      "direction": "고산들은 좌측 중경에서 두 손으로 창을 쥐고 우측 강영란을 향해 돌진 자세를 취하며, 창끝(흰 사슴뿔 촉)은 강영란의 가슴 높이에 거의 닿아 있다 — 촉 끝이 그녀의 흰 도포 가슴팍 앞에 접해 있고 가슴에는 검붉은 얼룩이 번져 있다. 다만 실제로 몸을 '꿰뚫고' 나온 상태는 아니고 접촉 직전/막 닿은 순간에 가깝다. 두 인물의 시선은 서로 맞물려 있고 창 자루가 프레임 좌하→중앙 대각선을 이룬다.",
      "built_space": "이전 스틸의 피 묻은 거석, 돌 제단, 화톳불 화로(좌·우), 제기 상, 어두운 숲이 그대로 유지 — 장소 락 일치. 제단은 강영란 뒤/좌측으로 물러나며 창의 연장선상에 위치. 인물 대비 배경 요소 스케일 자연스럽고 과장된 전경화 없음. 붉은 달은 프레임 밖이나 붉은 화광이 지배.",
      "entities": "여성: 백발 쪽머리, 60대 후반 한국 여성, 레퍼런스의 족두리(꽃장식 모자), 색동 소매단, 노리개, 얼굴의 상처·먹 자국까지 재현되어 캐릭터 레퍼런스와 잘 일치. 흰 도포의 가슴에 피 얼룩 있음. 남자: 갈색 비니, 올리브 작업 조끼, 청/네이비 체크 셔츠, 작업 바지 — 고산들 레퍼런스 의상과 정확히 일치, 40대 중반 각진 얼굴. 창은 나무 자루에 끈으로 감긴 흰 뿔촉으로 프롭 레퍼런스와 부합. 산들의 자상·총상 유혈 흔적은 뚜렷하지 않음.",
      "hard_violations": [],
      "physics": "남자는 앞발을 앞으로 내딛고 뒷발로 밀어내는 런지로 지면에 확실히 접지, 양손이 자루를 정확히 파지 — 돌진의 하중 이동이 읽힌다. 여자는 두 발로 지면에 서 있고 몸이 반쯤 돌아선 상태로 창을 맞이한다. 공중에 뜬 무지지 물체나 인체는 없다. 다만 관통 충격에 대한 몸의 반작용(밀림·경직)은 약하다."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": true,
    "readings": [
     {
      "label": "A",
      "direction": "고산들은 프레임 좌측에서 우측 강영란을 향해 몸을 낮추고 창을 겨눈다. 창끝(흰 사슴뿔)은 강영란의 가슴 높이를 향하지만 실제로는 그녀의 몸 앞 수십 센티 지점에서 멈춰 있고 관통하지 않는다. 두 사람의 시선은 서로 마주쳐 eye line은 성립. 촬영각은 측면 3/4, 가슴 높이에 가깝고 미디엄~미디엄 풀숏.",
      "built_space": "배경은 참조 스틸의 의식 공간과 일치: 피 묻은 거대한 바위, 그 앞의 낮은 석재 제단, 좌우 화롯불 두 곳, 삼각대 화로, 흙바닥, 어두운 숲. 제단은 창의 연장선상 뒤편에 놓여 앞면·측면이 후퇴하며 보인다. 구조물 중복이나 불가능한 요소는 없다.",
      "entities": "강영란: 백발을 단정히 틀어올림, 화관(족두리) 착용, 흰 한복에 색동 소매단 — 캐릭터 참조와 대체로 일치. 얼굴에 검은 글자 자국이 일부 보이는데 이는 참조에 없는 요소(경미한 이탈). 흰 저고리에 핏자국이 미미해 '피 묻은 흰 도포'는 약함. 고산들: 40대 한국 남성, 비니, 올리브 조끼, 체크 셔츠 — 캐릭터 참조와 일치. 창: 나무 자루에 흰 사슴뿔 촉, 참조와 유사.",
      "physics": "두 인물 모두 발이 땅에 붙어 있다. 고산들은 앞다리 굴곡·뒷다리 뻗음으로 돌진 자세를 유지하고 양손이 자루를 실제로 쥐고 있다. 공중에 떠 있는 물체는 없고 창은 두 손에 지지된다. 다만 관통 순간의 충격 물리(피, 옷 파열, 뒤로 밀리는 몸)는 전혀 없다.",
      "hard_violations": []
     },
     {
      "label": "B",
      "direction": "고산들 역할의 남자가 좌측에서 우측 강영란을 향해 창을 겨눈다. 창끝은 그녀 가슴보다 낮은 복부~허리 높이 허공에서 멈춰 있고 그녀 몸에 닿지 않는다. 서로 시선은 마주친다. 각도·스케일은 A와 사실상 동일.",
      "built_space": "동일한 의식 공간, 피 묻은 바위, 석재 제단, 화롯불, 삼각 화로. 구조 자체의 오류는 없다.",
      "entities": "고산들: 참조의 비니·올리브 조끼·체크 셔츠·현대 작업복이 전부 사라지고 상투와 망건, 흰 조선 저고리로 완전히 교체됨 — 지정된 인물의 의상 락 위반이며 현대 인물이 아닌 다른 시대 인물로 보인다. 강영란: 화관(족두리)이 사라졌고 색동 소매단·문양 도포가 없는 밋밋한 흰 한복, 혈흔도 거의 없음 — 참조 이탈. 창은 사슴뿔 촉이 참조보다 길고 금속 창날에 가까워 보인다.",
      "physics": "두 인물 모두 발이 지면에 있고 창은 양손에 쥐어져 있다. 부유물은 없다. 그러나 관통·충격의 흔적이 전혀 없어 샷 텍스트의 순간이 성립하지 않는다.",
      "hard_violations": [
       "지정 인물 고산들의 의상·헤어가 캐릭터 참조(현대 비니·작업복)에서 조선시대 상투·한복으로 전면 교체되어, 프레임 안의 인물이 계약이 지정한 인물로 읽히지 않음"
      ]
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "창끝이 강영란의 가슴을 '푹 꿰뚫은 순간'이 아니라 아직 닿지 않은 대치 상태라 핵심 액션은 미달이지만, 장소·조명·두 인물의 의상과 정체성(비니·조끼의 고산들, 족두리·색동 백발의 강영란)이 참조에 충실해 계약 손실이 적다."
     },
     {
      "label": "B",
      "score": 2,
      "verdict_ko": "관통 순간이 없는 것은 A와 같으면서, 계약이 요구하지 않은 변경 — 고산들을 상투·조선 한복 차림으로 바꾸고 강영란의 족두리·색동 소매·혈흔까지 지워버림 — 이 인물 정체성과 의상 락을 정면으로 깨뜨려 수리가 얻은 것보다 잃은 것이 훨씬 크다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "창끝이 강영란의 가슴을 '푹 꿰뚫은 순간'이 아니라 아직 닿지 않은 대치 상태라 핵심 액션은 미달이지만, 장소·조명·두 인물의 의상과 정체성(비니·조끼의 고산들, 족두리·색동 백발의 강영란)이 참조에 충실해 계약 손실이 적다."
     },
     {
      "label": "A",
      "score": 2,
      "verdict_ko": "관통 순간이 없는 것은 A와 같으면서, 계약이 요구하지 않은 변경 — 고산들을 상투·조선 한복 차림으로 바꾸고 강영란의 족두리·색동 소매·혈흔까지 지워버림 — 이 인물 정체성과 의상 락을 정면으로 깨뜨려 수리가 얻은 것보다 잃은 것이 훨씬 크다."
     }
    ],
    "all_candidates_fail": true,
    "readings": [
     {
      "label": "B",
      "direction": "고산들은 프레임 좌측에서 우측 강영란을 향해 몸을 낮추고 창을 겨눈다. 창끝(흰 사슴뿔)은 강영란의 가슴 높이를 향하지만 실제로는 그녀의 몸 앞 수십 센티 지점에서 멈춰 있고 관통하지 않는다. 두 사람의 시선은 서로 마주쳐 eye line은 성립. 촬영각은 측면 3/4, 가슴 높이에 가깝고 미디엄~미디엄 풀숏.",
      "built_space": "배경은 참조 스틸의 의식 공간과 일치: 피 묻은 거대한 바위, 그 앞의 낮은 석재 제단, 좌우 화롯불 두 곳, 삼각대 화로, 흙바닥, 어두운 숲. 제단은 창의 연장선상 뒤편에 놓여 앞면·측면이 후퇴하며 보인다. 구조물 중복이나 불가능한 요소는 없다.",
      "entities": "강영란: 백발을 단정히 틀어올림, 화관(족두리) 착용, 흰 한복에 색동 소매단 — 캐릭터 참조와 대체로 일치. 얼굴에 검은 글자 자국이 일부 보이는데 이는 참조에 없는 요소(경미한 이탈). 흰 저고리에 핏자국이 미미해 '피 묻은 흰 도포'는 약함. 고산들: 40대 한국 남성, 비니, 올리브 조끼, 체크 셔츠 — 캐릭터 참조와 일치. 창: 나무 자루에 흰 사슴뿔 촉, 참조와 유사.",
      "physics": "두 인물 모두 발이 땅에 붙어 있다. 고산들은 앞다리 굴곡·뒷다리 뻗음으로 돌진 자세를 유지하고 양손이 자루를 실제로 쥐고 있다. 공중에 떠 있는 물체는 없고 창은 두 손에 지지된다. 다만 관통 순간의 충격 물리(피, 옷 파열, 뒤로 밀리는 몸)는 전혀 없다.",
      "hard_violations": []
     },
     {
      "label": "A",
      "direction": "고산들 역할의 남자가 좌측에서 우측 강영란을 향해 창을 겨눈다. 창끝은 그녀 가슴보다 낮은 복부~허리 높이 허공에서 멈춰 있고 그녀 몸에 닿지 않는다. 서로 시선은 마주친다. 각도·스케일은 A와 사실상 동일.",
      "built_space": "동일한 의식 공간, 피 묻은 바위, 석재 제단, 화롯불, 삼각 화로. 구조 자체의 오류는 없다.",
      "entities": "고산들: 참조의 비니·올리브 조끼·체크 셔츠·현대 작업복이 전부 사라지고 상투와 망건, 흰 조선 저고리로 완전히 교체됨 — 지정된 인물의 의상 락 위반이며 현대 인물이 아닌 다른 시대 인물로 보인다. 강영란: 화관(족두리)이 사라졌고 색동 소매단·문양 도포가 없는 밋밋한 흰 한복, 혈흔도 거의 없음 — 참조 이탈. 창은 사슴뿔 촉이 참조보다 길고 금속 창날에 가까워 보인다.",
      "physics": "두 인물 모두 발이 지면에 있고 창은 양손에 쥐어져 있다. 부유물은 없다. 그러나 관통·충격의 흔적이 전혀 없어 샷 텍스트의 순간이 성립하지 않는다.",
      "hard_violations": [
       "지정 인물 고산들의 의상·헤어가 캐릭터 참조(현대 비니·작업복)에서 조선시대 상투·한복으로 전면 교체되어, 프레임 안의 인물이 계약이 지정한 인물로 읽히지 않음"
      ]
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 5,
     "B": 11
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": true,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "needs_reshoot": true,
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S110sh21"
  }
 },
 "S110sh41::bgfirst_bg": {
  "input_fingerprint": "6aa2a20a95204e5f",
  "prompt": "Turn the attached storyboard sketch into a photorealistic film still of\nits LOCATION ONLY, keeping the figures exactly as they are.\n\nKEEP EXACTLY: the camera framing, the horizon, and where every element\nsits in the frame. Each mannequin figure stays a plain grey featureless\nmannequin standing in the very same spot, at the same size, in the same\npose, turned the same way — do not turn them into people, do not move,\nrotate, mirror or re-pose them, do not add or remove figures.\n\nBUILD PHOTOREALISTICALLY: everything that is not a figure — ground,\nsurfacing, structures, vegetation, sky, water, distance. The bare lines\nof the sketch are a layout guide; replace them with the real materials,\ndepth and lighting of the place described below.\n\nSHOT TEXT this background must serve (Korean): 붉은 달빛 아래 육혈귀들이 서로를 뜯어먹는 핏빛 지옥도 한가운데서, 온몸이 피범벅이 된 김지후(강수리영)가 차분한 표정으로 참상을 내려다보는 꼿꼿한 전신.\n\nLOCATION (lock): At the center of the blood-soaked ritual clearing beneath the red eclipsed moon.\n\nTIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Settle the backward crane at a slightly elevated wide position, preserving 김지후(강수리영)'s full body at the visual center while violent clusters occupy the lower foreground and both sides. She remains vertically composed and still amid circling diagonals of bodies, with the altar and red moon anchoring the deeper ritual space rather than replacing the carnage as the focus.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-center of the frame, midground; 금월도 마을 주민 (빙의 상태) in the lower-left of the frame, foreground; altar and surrounding carnage in the lower-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 서로를 공격하는 육혈귀들과 시체 더미 (서로 물어뜯고 할퀴며 죽어가는 상태); used as Foreground and side-plane action surrounding the calm central figure; 중앙 제단과 거대한 바위 (광란의 중심에 남아 있는 상태) — Their front-facing ritual side sits behind the central figure in the elevated view; used as Central ritual-space anchor behind 김지후(강수리영); 붉은 보름달 (붉은빛을 드러낸 상태); used as Upper-background celestial anchor establishing the red world described in the scene.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nTHINGS AT THIS PLACE:\nThe site is organized along a forest path that enters from the outer forest zone and proceeds inward through a dense stand of pine trunks. At its inner end, the recognizable path disappears into a trackless deep-forest interior. Beyond and around this transition, several stone markers inscribed with Suri-young's name are distributed at separate points along the continuing walking line. Dense pine trunks enclose the route on both sides and continue past the path disappearance point. Undergrowth fills the spaces between the trunks, borders the path, and occupies the off-path ground around the dispersed marker stones.\n- path: Outer-to-deep forest path\n- path end: Forest path disappearance point\n- marker stone: Group of stone markers inscribed with Suri-young's name\n- tree stand: Dense pine trunk stand\n- undergrowth: Intervening forest undergrowth\n\nWORLD FACTS (creator-confirmed — always true):\n- Region (real-world reference): 대한민국 서울권과 서남해안 일대, 대한민국 영토 내 서해의 가상 외딴섬 공동체\n- Era: 주 배경은 스마트폰과 현대 경찰·교통 체계가 존재하는 2020년대 중반의 현대이며, 1670년 조선시대와 20세기 전반 일제강점기 전후의 역사적 재현 구간이 명확히 구분되어 병존한다.\n- 현대 구간은 2020년대 대한민국의 생활상으로, 1670년대 구간은 조선 후기의 복식과 생활 도구로, 20세기 전반 구간은 당시 조선인 복식과 일본군 장비로 표현한다. 시대 전환이 없는 프레임에는 서로 다른 시대의 의상·무기·교통수단·전자기기를 혼합하지 않는다.\n- 현대 육지에서는 스마트폰, 메신저 앱, 지도 앱, CCTV, 바코드 출근 시스템, 현대 경찰 장비, 승용차·구형 SUV·동력 어선이 자연스럽게 존재한다. 섬에서는 휴대전화 신호와 인터넷 연결이 불안정하거나 차단되며, 제한적인 전기 조명과 오래된 생활 설비를 우선하되 현대 물품 자체가 전혀 없는 원시사회로 오해하지 않는다. 역사 구간에는 디지털 기기와 현대 차량을 배제한다.\n\nTHINGS THAT LIVE AT THIS PLACE (what they are, not where they go): the\nlist above names what permanently belongs to this site. Where each one\nsits in this frame is decided by the attached sketch alone — do not add\nor reposition anything on the strength of the list, build only what the\nsketch already has lines for, and do not invent facilities that are not\nlisted.\n\nNo text, no captions, no diagram symbols, no markers anywhere in the\nimage.",
  "effective_prompt": "Turn the attached storyboard sketch into a photorealistic film still of\nits LOCATION ONLY, keeping the figures exactly as they are.\n\nKEEP EXACTLY: the camera framing, the horizon, and where every element\nsits in the frame. Each mannequin figure stays a plain grey featureless\nmannequin standing in the very same spot, at the same size, in the same\npose, turned the same way — do not turn them into people, do not move,\nrotate, mirror or re-pose them, do not add or remove figures.\n\nBUILD PHOTOREALISTICALLY: everything that is not a figure — ground,\nsurfacing, structures, vegetation, sky, water, distance. The bare lines\nof the sketch are a layout guide; replace them with the real materials,\ndepth and lighting of the place described below.\n\nSHOT TEXT this background must serve (Korean): 붉은 달빛 아래 육혈귀들이 서로를 뜯어먹는 핏빛 지옥도 한가운데서, 온몸이 피범벅이 된 김지후(강수리영)가 차분한 표정으로 참상을 내려다보는 꼿꼿한 전신.\n\nLOCATION (lock): At the center of the blood-soaked ritual clearing beneath the red eclipsed moon.\n\nTIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Settle the backward crane at a slightly elevated wide position, preserving 김지후(강수리영)'s full body at the visual center while violent clusters occupy the lower foreground and both sides. She remains vertically composed and still amid circling diagonals of bodies, with the altar and red moon anchoring the deeper ritual space rather than replacing the carnage as the focus.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-center of the frame, midground; 금월도 마을 주민 (빙의 상태) in the lower-left of the frame, foreground; altar and surrounding carnage in the lower-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 서로를 공격하는 육혈귀들과 시체 더미 (서로 물어뜯고 할퀴며 죽어가는 상태); used as Foreground and side-plane action surrounding the calm central figure; 중앙 제단과 거대한 바위 (광란의 중심에 남아 있는 상태) — Their front-facing ritual side sits behind the central figure in the elevated view; used as Central ritual-space anchor behind 김지후(강수리영); 붉은 보름달 (붉은빛을 드러낸 상태); used as Upper-background celestial anchor establishing the red world described in the scene.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nTHINGS AT THIS PLACE:\nThe site is organized along a forest path that enters from the outer forest zone and proceeds inward through a dense stand of pine trunks. At its inner end, the recognizable path disappears into a trackless deep-forest interior. Beyond and around this transition, several stone markers inscribed with Suri-young's name are distributed at separate points along the continuing walking line. Dense pine trunks enclose the route on both sides and continue past the path disappearance point. Undergrowth fills the spaces between the trunks, borders the path, and occupies the off-path ground around the dispersed marker stones.\n- path: Outer-to-deep forest path\n- path end: Forest path disappearance point\n- marker stone: Group of stone markers inscribed with Suri-young's name\n- tree stand: Dense pine trunk stand\n- undergrowth: Intervening forest undergrowth\n\nWORLD FACTS (creator-confirmed — always true):\n- Region (real-world reference): 대한민국 서울권과 서남해안 일대, 대한민국 영토 내 서해의 가상 외딴섬 공동체\n- Era: 주 배경은 스마트폰과 현대 경찰·교통 체계가 존재하는 2020년대 중반의 현대이며, 1670년 조선시대와 20세기 전반 일제강점기 전후의 역사적 재현 구간이 명확히 구분되어 병존한다.\n- 현대 구간은 2020년대 대한민국의 생활상으로, 1670년대 구간은 조선 후기의 복식과 생활 도구로, 20세기 전반 구간은 당시 조선인 복식과 일본군 장비로 표현한다. 시대 전환이 없는 프레임에는 서로 다른 시대의 의상·무기·교통수단·전자기기를 혼합하지 않는다.\n- 현대 육지에서는 스마트폰, 메신저 앱, 지도 앱, CCTV, 바코드 출근 시스템, 현대 경찰 장비, 승용차·구형 SUV·동력 어선이 자연스럽게 존재한다. 섬에서는 휴대전화 신호와 인터넷 연결이 불안정하거나 차단되며, 제한적인 전기 조명과 오래된 생활 설비를 우선하되 현대 물품 자체가 전혀 없는 원시사회로 오해하지 않는다. 역사 구간에는 디지털 기기와 현대 차량을 배제한다.\n\nTHINGS THAT LIVE AT THIS PLACE (what they are, not where they go): the\nlist above names what permanently belongs to this site. Where each one\nsits in this frame is decided by the attached sketch alone — do not add\nor reposition anything on the strength of the list, build only what the\nsketch already has lines for, and do not invent facilities that are not\nlisted.\n\nNo text, no captions, no diagram symbols, no markers anywhere in the\nimage.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S110sh41__bgfirst_bg.png",
  "asset_id": "d1ef11f2-665b-4a0f-bb56-a5df7b6d7fd2",
  "input_asset_ids": [
   "5958c505-c07a-48c0-b981-01c1c624543d"
  ]
 },
 "S110sh41": {
  "input_fingerprint": "b0be863183e411db",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nSHOT TEXT (authoritative, Korean): 붉은 달빛 아래 육혈귀들이 서로를 뜯어먹는 핏빛 지옥도 한가운데서, 온몸이 피범벅이 된 김지후(강수리영)가 차분한 표정으로 참상을 내려다보는 꼿꼿한 전신.\n\nLOCATION (lock): At the center of the blood-soaked ritual clearing beneath the red eclipsed moon. The shot takes place here — the attached STORYBOARD SKETCH fixes the staging, camera and figure placement of this exact place. No location photograph is attached — build the location itself strictly from the location text above and the shot text, inventing nothing beyond them.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran and the headless Sandeul remain dead at the ritual ground amid the spreading corpses and blood. Jihu remains possessed, blood-covered, marked with black script, and golden-eyed.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Replace every grey or outlined mannequin figure in the FIRST attached\nimage with the real people the CHARACTER REFERENCE images show, and\noutput a photorealistic film still. Each mannequin becomes the character\nwhose POSE CANON and the shot text match that figure's pose and\nposition; never repeat one person across two mannequins. If no CHARACTER\nREFERENCE image is attached for a figure, still replace it with a\nplausible real person consistent with the text. Every mannequin becomes\na person — erasing a figure, or leaving one out, is not a replacement,\nand the number of figures never changes.\n\nKEEP EXACTLY: the background of the first image, the camera framing,\nand each mannequin's position, size, pose and the direction it is\nturned. A person must be exactly where their mannequin was, at the same\nscale, in the same pose, facing the same way — never mirrored, never\nre-staged, never straightened up, stood upright or otherwise re-posed.\nNo mannequin, grey figure or sketch line may remain anywhere in the\noutput.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nSHOT TEXT (authoritative, Korean): 붉은 달빛 아래 육혈귀들이 서로를 뜯어먹는 핏빛 지옥도 한가운데서, 온몸이 피범벅이 된 김지후(강수리영)가 차분한 표정으로 참상을 내려다보는 꼿꼿한 전신.\n\nLOCATION (lock): At the center of the blood-soaked ritual clearing beneath the red eclipsed moon. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran and the headless Sandeul remain dead at the ritual ground amid the spreading corpses and blood. Jihu remains possessed, blood-covered, marked with black script, and golden-eyed.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Replace every grey or outlined mannequin figure in the FIRST attached\nimage with the real people the CHARACTER REFERENCE images show, and\noutput a photorealistic film still. Each mannequin becomes the character\nwhose POSE CANON and the shot text match that figure's pose and\nposition; never repeat one person across two mannequins. If no CHARACTER\nREFERENCE image is attached for a figure, still replace it with a\nplausible real person consistent with the text. Every mannequin becomes\na person — erasing a figure, or leaving one out, is not a replacement,\nand the number of figures never changes.\n\nKEEP EXACTLY: the background of the first image, the camera framing,\nand each mannequin's position, size, pose and the direction it is\nturned. A person must be exactly where their mannequin was, at the same\nscale, in the same pose, facing the same way — never mirrored, never\nre-staged, never straightened up, stood upright or otherwise re-posed.\nNo mannequin, grey figure or sketch line may remain anywhere in the\noutput.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nSHOT TEXT (authoritative, Korean): 붉은 달빛 아래 육혈귀들이 서로를 뜯어먹는 핏빛 지옥도 한가운데서, 온몸이 피범벅이 된 김지후(강수리영)가 차분한 표정으로 참상을 내려다보는 꼿꼿한 전신.\n\nLOCATION (lock): At the center of the blood-soaked ritual clearing beneath the red eclipsed moon. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran and the headless Sandeul remain dead at the ritual ground amid the spreading corpses and blood. Jihu remains possessed, blood-covered, marked with black script, and golden-eyed.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "C": "Replace every grey or outlined mannequin figure in the FIRST attached\nimage with the real people the CHARACTER REFERENCE images show, and\noutput a photorealistic film still. Each mannequin becomes the character\nwhose POSE CANON and the shot text match that figure's pose and\nposition; never repeat one person across two mannequins. If no CHARACTER\nREFERENCE image is attached for a figure, still replace it with a\nplausible real person consistent with the text. Every mannequin becomes\na person — erasing a figure, or leaving one out, is not a replacement,\nand the number of figures never changes.\n\nKEEP EXACTLY: the background of the first image, the camera framing,\nand each mannequin's position, size, pose and the direction it is\nturned. A person must be exactly where their mannequin was, at the same\nscale, in the same pose, facing the same way — never mirrored, never\nre-staged, never straightened up, stood upright or otherwise re-posed.\nNo mannequin, grey figure or sketch line may remain anywhere in the\noutput.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, total lunar eclipse and blood-red moon.\n\nSHOT TEXT (authoritative, Korean): 붉은 달빛 아래 육혈귀들이 서로를 뜯어먹는 핏빛 지옥도 한가운데서, 온몸이 피범벅이 된 김지후(강수리영)가 차분한 표정으로 참상을 내려다보는 꼿꼿한 전신.\n\nLOCATION (lock): At the center of the blood-soaked ritual clearing beneath the red eclipsed moon. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran and the headless Sandeul remain dead at the ritual ground amid the spreading corpses and blood. Jihu remains possessed, blood-covered, marked with black script, and golden-eyed.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S110sh41__bgfirst_bg.png"
    }
   ],
   "B": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S110sh41__bgfirst_bg.png"
    }
   ],
   "C": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S110sh41__bgfirst_bg.png"
    }
   ]
  },
  "totals": {
   "B": 7,
   "A": 4,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "스토리보드 구도를 충실히 따르며, 온몸이 피범벅이 된 여성 캐릭터와 서로 물어뜯는 육혈귀들의 핏빛 액션을 가장 정확하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "주변 인물들의 액션과 유혈 표현은 훌륭하나, 중앙에 지정된 여성 배우 대신 남성 캐릭터를 배치하여 정체성 지시를 위반했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "중앙 인물이 피범벅 상태가 아니며, 주변 인물들 역시 서로 물어뜯는 대신 단순히 붙잡고 있는 모습으로 묘사되어 연출 지시를 크게 벗어났습니다."
   }
  ],
  "refs": [
   {
    "label": "SHOT BACKGROUND",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S110sh41__bgfirst_bg.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "중앙에 서 있는 여성(지후)의 피부와 옷에 지시문에 명시된 검은색 문자(black script)가 전혀 없습니다.",
     "fix_en": "Paint black script markings onto the face, limbs, and dress of the woman standing in the center.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "지시문에 '머리 없는 산들'이 명시되어 있으나, 우측 전경에서 뜯어먹히고 있는 시체의 머리가 온전히 남아있습니다.",
     "fix_en": "Erase the head of the victim on the ground in the right foreground to leave a bloody neck stump, adjusting the attacker's hands to hold the neck or shoulders.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Paint black script markings onto the face, limbs, and dress of the woman standing in the center.\n- Erase the head of the victim on the ground in the right foreground to leave a bloody neck stump, adjusting the attacker's hands to hold the neck or shoulders.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S110sh41__bgfirst_bg.png",
   "bg_asset_id": "d1ef11f2-665b-4a0f-bb56-a5df7b6d7fd2",
   "bg_record_key": "S110sh41::bgfirst_bg",
   "chain_winner": true,
   "authority": "lane_conti_only"
  },
  "ref_mode": "lane(map_marker): 스케치+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S110sh25"
  }
 },
 "S111sh11::bgfirst_bg": {
  "input_fingerprint": "e685140a71df0a26",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 정인우가 쥔 사슴뿔 목걸이가 김지후(강수리영)의 가슴팍에 깊숙이 꽂혀 힘이 가해지는 순간의 클로즈업.\n\nLOCATION (lock): Among piled bodies and blood pools in the forest ritual clearing after the massacre.\n\nTIME OF DAY (lock): night, clear full moon after eclipse.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Begin the dolly-out at chest height in a tight close frame, centering the point where 정인우's hand drives the sharp antler necklace into 김지후(강수리영)'s chest. His forearm enters diagonally from the lower-left as her torso recoils toward the upper-right, leaving enough of both bodies visible to make the force and ownership of the strike unambiguous.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 날카로운 흰색 사슴뿔 목걸이 (가슴팍에 깊숙이 꽂혀 힘이 가해지는 상태) — Its pointed end is directed from the lower-left into the front of 김지후(강수리영)'s chest; used as Central impact object linking 정인우's hand to 김지후(강수리영)'s torso; 시체 더미와 흐르는 피 (시체가 쌓이고 피가 좁은 내와 웅덩이를 이룬 상태); used as Blurred rear-plane evidence of the massacre beyond the impact.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 정인우가 쥔 사슴뿔 목걸이가 김지후(강수리영)의 가슴팍에 깊숙이 꽂혀 힘이 가해지는 순간의 클로즈업.\n\nLOCATION (lock): Among piled bodies and blood pools in the forest ritual clearing after the massacre.\n\nTIME OF DAY (lock): night, clear full moon after eclipse.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Begin the dolly-out at chest height in a tight close frame, centering the point where 정인우's hand drives the sharp antler necklace into 김지후(강수리영)'s chest. His forearm enters diagonally from the lower-left as her torso recoils toward the upper-right, leaving enough of both bodies visible to make the force and ownership of the strike unambiguous.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 날카로운 흰색 사슴뿔 목걸이 (가슴팍에 깊숙이 꽂혀 힘이 가해지는 상태) — Its pointed end is directed from the lower-left into the front of 김지후(강수리영)'s chest; used as Central impact object linking 정인우's hand to 김지후(강수리영)'s torso; 시체 더미와 흐르는 피 (시체가 쌓이고 피가 좁은 내와 웅덩이를 이룬 상태); used as Blurred rear-plane evidence of the massacre beyond the impact.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S111sh11__bgfirst_bg.png",
  "asset_id": "ca00d3a1-6f84-411d-b901-67bbcc8e87c9",
  "input_asset_ids": [
   "fd6456a4-1940-4005-bcda-6311c4541b29",
   "5f354400-a374-4a5d-bf4a-b9e057f77e78"
  ]
 },
 "S111sh11": {
  "input_fingerprint": "c00713cbdcb8d50c",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, clear full moon after eclipse.\n\nSHOT TEXT (authoritative, Korean): 정인우가 쥔 사슴뿔 목걸이가 김지후(강수리영)의 가슴팍에 깊숙이 꽂혀 힘이 가해지는 순간의 클로즈업.\n\nLOCATION (lock): Among piled bodies and blood pools in the forest ritual clearing after the massacre. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The ritual ground remains covered with corpses, flowing blood, and pools; Youngran and Sandeul remain dead. Jihu remains marked and golden-eyed as Inwoo drives her recovered antler-bone pendant into her chest.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, clear full moon after eclipse.\n\nSHOT TEXT (authoritative, Korean): 정인우가 쥔 사슴뿔 목걸이가 김지후(강수리영)의 가슴팍에 깊숙이 꽂혀 힘이 가해지는 순간의 클로즈업.\n\nLOCATION (lock): Among piled bodies and blood pools in the forest ritual clearing after the massacre. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The ritual ground remains covered with corpses, flowing blood, and pools; Youngran and Sandeul remain dead. Jihu remains marked and golden-eyed as Inwoo drives her recovered antler-bone pendant into her chest.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, clear full moon after eclipse.\n\nSHOT TEXT (authoritative, Korean): 정인우가 쥔 사슴뿔 목걸이가 김지후(강수리영)의 가슴팍에 깊숙이 꽂혀 힘이 가해지는 순간의 클로즈업.\n\nLOCATION (lock): Among piled bodies and blood pools in the forest ritual clearing after the massacre. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The ritual ground remains covered with corpses, flowing blood, and pools; Youngran and Sandeul remain dead. Jihu remains marked and golden-eyed as Inwoo drives her recovered antler-bone pendant into her chest.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S111sh11__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S111sh11.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L27B03.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "클로즈업 샷 크기, 사슴뿔 목걸이로 가슴을 찌르는 동작, 금빛 눈동자 등 프롬프트의 요구사항을 정확히 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "클로즈업 샷이 아니며, 사슴뿔 목걸이나 가슴을 찌르는 동작, 금빛 눈동자 등의 핵심 디테일이 누락되었습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "클로즈업 샷 크기, 사슴뿔 목걸이로 가슴을 찌르는 동작, 금빛 눈동자 등 프롬프트의 요구사항을 정확히 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "클로즈업 샷이 아니며, 사슴뿔 목걸이나 가슴을 찌르는 동작, 금빛 눈동자 등의 핵심 디테일이 누락되었습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "사슴뿔 목걸이를 가슴에 꽂는 핵심 액션과 클로즈업 샷, 금빛 눈동자 및 배경의 시체 등 캐리드 스테이트를 충실히 구현했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "사슴뿔이 존재하지 않아 핵심 액션이 누락되었으며, 샷 크기가 너무 넓고 금빛 눈동자나 배경의 시체 등 세부 설정이 모두 빠졌습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "사슴뿔 목걸이를 가슴에 꽂는 핵심 액션과 클로즈업 샷, 금빛 눈동자 및 배경의 시체 등 캐리드 스테이트를 충실히 구현했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "사슴뿔이 존재하지 않아 핵심 액션이 누락되었으며, 샷 크기가 너무 넓고 금빛 눈동자나 배경의 시체 등 세부 설정이 모두 빠졌습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 6
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 15,
   "B": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "클로즈업 샷 크기, 사슴뿔 목걸이로 가슴을 찌르는 동작, 금빛 눈동자 등 프롬프트의 요구사항을 정확히 구현했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "클로즈업 샷이 아니며, 사슴뿔 목걸이나 가슴을 찌르는 동작, 금빛 눈동자 등의 핵심 디테일이 누락되었습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L27B03.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "'스케치의 배경 선을 무시하라'는 지시를 어기고 스케치에 있던 탁자와 시체들을 그대로 반영하여, 원래 배경의 돌담 위에 탁자가 공중에 떠 있고 원근법이 완전히 훼손되었습니다.",
     "fix_en": "Remove the floating table, lantern, and out-of-perspective bodies from the background area between the characters, perfectly restoring the original stone wall, ground, and items exactly as they appear in the SHOT BACKGROUND.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "사슴뿔을 쥐고 있는 남자의 오른손 손가락 구조가 기형적이며 뿔과 융합되어 있습니다.",
     "fix_en": "Redraw the man's right hand gripping the antler to have anatomically correct fingers that cleanly wrap around the bone without merging into it.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the floating table, lantern, and out-of-perspective bodies from the background area between the characters, perfectly restoring the original stone wall, ground, and items exactly as they appear in the SHOT BACKGROUND.\n- Redraw the man's right hand gripping the antler to have anatomically correct fingers that cleanly wrap around the bone without merging into it.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S111sh11__bgfirst_bg.png",
   "bg_asset_id": "ca00d3a1-6f84-411d-b901-67bbcc8e87c9",
   "bg_record_key": "S111sh11::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S111sh14": {
  "input_fingerprint": "8ac002eb90998a6d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, clear full moon after eclipse.\n\nSHOT TEXT (authoritative, Korean): 무릎을 꿇은 김지후(강수리영)의 몸에서 그분(검은 연기 형체)이 짙은 연기 형태로 스멀스멀 빠져나오는 찰나.\n\nLOCATION (lock): On the corpse-strewn ground of the ritual clearing under restored moonlight. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the corpse-strewn ritual clearing, moonlight, blood pools, and kneeling woman's position. Exclude the pendant's impact close-up and show black smoke leaving her body.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu kneels with the pendant wound in her chest as the black entity leaves her body; the antler pendant simultaneously crumbles into ash. The surrounding corpses remain where they fell.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 7,
   "C": 5,
   "A": 4
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "카메라 앵글을 전면으로 변경하라는 지시를 잘 따랐으며, 이전 샷에서 확립된 위를 향한 자세와 가슴 상처에서 빠져나오는 검은 연기를 충실히 구현함."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "카메라 구도는 변경했으나 레퍼런스에서 확립된 시선(위를 보는 자세)을 임의로 아래로 바꾸었으며, 뿔 펜던트가 금속 재질처럼 왜곡됨."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "레퍼런스 이미지의 카메라 구도(어깨 너머 샷)를 절대 복사하지 말라는 제한 지침을 위반하고 프레임을 거의 동일하게 유지함."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S111sh11_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "가슴에 박힌 채 재로 부서져야 할 사슴 뿔 펜던트가 바닥에 온전한 모양으로 놓여 있습니다.",
     "fix_en": "Remove the intact antler pendant from the ground in front of her and draw it sticking out of her chest wound, visibly crumbling into ash.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "이전 샷에서 상체를 뒤로 젖히고 고개를 든 자세가 정확히 유지되어야 하지만, 상체를 꼿꼿이 세우고 앉은 자세로 바뀌었습니다.",
     "fix_en": "Redraw her upper body and head to lean backward with her chin raised, matching the exact posture from the reference image.",
     "needs_regeneration": false,
     "unfixable": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the intact antler pendant from the ground in front of her and draw it sticking out of her chest wound, visibly crumbling into ash.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S111sh11"
  }
 },
 "S111sh19": {
  "input_fingerprint": "c950237fe7cbe9cf",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): night, clear full moon after eclipse.\n\nSHOT TEXT (authoritative, Korean): 정인우가 참혹한 슬픔에 잠겨 굵은 눈물을 뚝뚝 흘리며 오열하는 얼굴 클로즈업.\n\nLOCATION (lock): Beside the fallen woman amid the devastated ritual clearing and its bodies. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the moonlit clearing, heaps of bodies, blood pools, and the fallen woman's nearby position. Exclude the smoke-extraction emphasis and focus on the surviving man's grief.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu lies collapsed after the entity and pendant have vanished; the ritual ground remains littered with corpses and blood.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 8,
   "A": 7,
   "B": 4
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "요구된 '얼굴 클로즈업' 프레이밍을 가장 정확히 구현했으며, 기이한 형체가 사라지고 쓰러진 여성의 디테일과 오열하는 표정 연기가 지시문과 완벽히 일치합니다."
   },
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지시된 감정과 주변 상황은 충실하게 묘사되었으나, 얼굴 클로즈업이라기에는 카메라가 다소 물러난 미디엄 클로즈업에 가까워 C보다 프레이밍 우선순위에서 밀립니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "턱에서 떨어지는 눈물이 물리적으로 불가능한 고체 가닥처럼 렌더링되었으며, 완전히 사라져야 할 뿔 형태의 물체가 여성 옆 바닥에 그대로 남아있어 설정에 위배됩니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S111sh14_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "배경에 쓰러진 시신들의 배치와 자세가 이전 장면(레퍼런스)과 전혀 일치하지 않습니다.",
     "fix_en": "The layout and poses of the background corpses fail to match the locked reference image.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "왼쪽 배경에 쓰러진 여성의 상의에 레퍼런스에는 없는 붉은색 무늬가 생겼습니다.",
     "fix_en": "Remove the red patterns from the collapsed woman's top, restoring it to the plain fabric seen in the reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the red patterns from the collapsed woman's top, restoring it to the plain fabric seen in the reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S111sh14"
  }
 },
 "S112sh6::bgfirst_bg": {
  "input_fingerprint": "6bd94ba1a076eac9",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 낚시 어선 갑판에 먼저 올라탄 정인우가 선착장에 선 김지후(강수리영)를 향해 손을 뻗어 내민 모습.\n\nLOCATION (lock): At the edge between the island pier and the deck of a moored fishing boat in daylight.\n\nTIME OF DAY (lock): day, clear sky.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height beside the gap, hold a static, slightly wide view with 정인우 on the deck at frame left and 김지후(강수리영) on the pier at frame right. Their extended arms form a diagonal toward the center without touching, while the visible separation beneath their hands preserves the physical choice still between them.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: gap between the deck and pier beneath the offered hands in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: fishing boat deck (정인우 is aboard) — The deck side carrying 정인우 faces diagonally toward the camera; used as Supports 정인우 on the left and establishes that he has already boarded; pier (김지후(강수리영) remains on it) — The pier surface carrying 김지후(강수리영) is visible across the central gap; used as Supports 김지후(강수리영) on the right and establishes her remaining position ashore; sea (dazzling and sparkling) — The visible water surface catches daylight in bright glints between the boat and pier; used as Appears below and beyond the offered hands, visually maintaining their separation; blue sky (clear and sunny); used as Provides the open upper background above both figures.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 낚시 어선 갑판에 먼저 올라탄 정인우가 선착장에 선 김지후(강수리영)를 향해 손을 뻗어 내민 모습.\n\nLOCATION (lock): At the edge between the island pier and the deck of a moored fishing boat in daylight.\n\nTIME OF DAY (lock): day, clear sky.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height beside the gap, hold a static, slightly wide view with 정인우 on the deck at frame left and 김지후(강수리영) on the pier at frame right. Their extended arms form a diagonal toward the center without touching, while the visible separation beneath their hands preserves the physical choice still between them.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: gap between the deck and pier beneath the offered hands in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: fishing boat deck (정인우 is aboard) — The deck side carrying 정인우 faces diagonally toward the camera; used as Supports 정인우 on the left and establishes that he has already boarded; pier (김지후(강수리영) remains on it) — The pier surface carrying 김지후(강수리영) is visible across the central gap; used as Supports 김지후(강수리영) on the right and establishes her remaining position ashore; sea (dazzling and sparkling) — The visible water surface catches daylight in bright glints between the boat and pier; used as Appears below and beyond the offered hands, visually maintaining their separation; blue sky (clear and sunny); used as Provides the open upper background above both figures.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S112sh6__bgfirst_bg.png",
  "asset_id": "9e9e53d3-32aa-43d6-934f-50ba0c828bbb",
  "input_asset_ids": [
   "4f27bc1f-bea7-47ff-8122-7e696327e1c7",
   "ab7bf9fc-815c-4d28-aec7-47d7103bc8d7",
   "6434b544-0a16-42ea-a909-412c18868a98"
  ]
 },
 "S112sh6": {
  "input_fingerprint": "b46d45b8ed2df133",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, clear sky.\n\nSHOT TEXT (authoritative, Korean): 낚시 어선 갑판에 먼저 올라탄 정인우가 선착장에 선 김지후(강수리영)를 향해 손을 뻗어 내민 모습.\n\nLOCATION (lock): At the edge between the island pier and the deck of a moored fishing boat in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant is permanently gone after turning to ash; both survivors retain the aftermath of the ritual ordeal.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, clear sky.\n\nSHOT TEXT (authoritative, Korean): 낚시 어선 갑판에 먼저 올라탄 정인우가 선착장에 선 김지후(강수리영)를 향해 손을 뻗어 내민 모습.\n\nLOCATION (lock): At the edge between the island pier and the deck of a moored fishing boat in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant is permanently gone after turning to ash; both survivors retain the aftermath of the ritual ordeal.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, clear sky.\n\nSHOT TEXT (authoritative, Korean): 낚시 어선 갑판에 먼저 올라탄 정인우가 선착장에 선 김지후(강수리영)를 향해 손을 뻗어 내민 모습.\n\nLOCATION (lock): At the edge between the island pier and the deck of a moored fishing boat in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s antler-bone pendant is permanently gone after turning to ash; both survivors retain the aftermath of the ritual ordeal.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S112sh6__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S112sh6.png"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L25B02.png"
    },
    {
     "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_slaughter_complex_sel.png"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "남자가 배에 먼저 올라타 부두에 있는 여자를 향해 손을 뻗는 동작과 위치 관계가 레퍼런스 및 샷 텍스트의 의도에 정확하게 부합합니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "배와 부두 사이의 거리가 비현실적으로 멀고 남자가 서 있는 구조물이 허공에 떠 있어 물리적으로 불가능한 공간 연출입니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "남자가 배에 먼저 올라타 부두에 있는 여자를 향해 손을 뻗는 동작과 위치 관계가 레퍼런스 및 샷 텍스트의 의도에 정확하게 부합합니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "배와 부두 사이의 거리가 비현실적으로 멀고 남자가 서 있는 구조물이 허공에 떠 있어 물리적으로 불가능한 공간 연출입니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "배에 올라탄 인물과 선착장에 선 인물이 서로 손을 뻗는 모습을 주어진 배경에 맞춰 물리적 오류 없이 정확하게 연출했습니다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "선착장에 서 있어야 할 여성 캐릭터의 하반신이 콘크리트 구조물 아래로 파묻힌 듯한 물리적으로 불가능한 위치에 있어 심각한 구조적 오류(Hard Violation)가 발생했습니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "배에 올라탄 인물과 선착장에 선 인물이 서로 손을 뻗는 모습을 주어진 배경에 맞춰 물리적 오류 없이 정확하게 연출했습니다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "선착장에 서 있어야 할 여성 캐릭터의 하반신이 콘크리트 구조물 아래로 파묻힌 듯한 물리적으로 불가능한 위치에 있어 심각한 구조적 오류(Hard Violation)가 발생했습니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 10,
     "B": 10
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 10,
   "B": 10
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "남자가 배에 먼저 올라타 부두에 있는 여자를 향해 손을 뻗는 동작과 위치 관계가 레퍼런스 및 샷 텍스트의 의도에 정확하게 부합합니다."
   },
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "배와 부두 사이의 거리가 비현실적으로 멀고 남자가 서 있는 구조물이 허공에 떠 있어 물리적으로 불가능한 공간 연출입니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L25B02.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_slaughter_complex_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "스케치의 배경 선을 무시하라는 지시를 위반하여, 원본 배경 이미지에는 없는 철제 발판과 난간을 배 위에 임의로 생성했습니다.",
     "fix_en": "Remove the invented metal platform and railing on the boat, restoring the original deck and blue crate from the background reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the invented metal platform and railing on the boat, restoring the original deck and blue crate from the background reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S112sh6__bgfirst_bg.png",
   "bg_asset_id": "9e9e53d3-32aa-43d6-934f-50ba0c828bbb",
   "bg_record_key": "S112sh6::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate",
   "seed_attached": true
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S112sh8": {
  "input_fingerprint": "66e571d4275751a5",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, clear sky.\n\nSHOT TEXT (authoritative, Korean): 무언가 결심한 듯 뻗었던 팔을 아래로 떨어뜨려 내린 채 굳은 표정을 짓고 있는 김지후(강수리영)의 모습.\n\nLOCATION (lock): On the island pier immediately beside the moored fishing boat's boarding point. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the sunny pier, fishing boat deck, blue sea, and the man's outstretched hand. Exclude the woman's reaching action and show her lowering her arm after deciding to stay.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s pendant remains gone, and she stays on the pier rather than boarding the boat.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "B": 8,
   "A": 4,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "새로운 카메라 앵글을 통해 굳은 표정과 팔을 내린 동작에 집중하여 지시사항을 훌륭하게 구현했습니다."
   },
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "레퍼런스 이미지의 카메라 구도를 그대로 복사하여 '카메라 프레이밍을 복사하지 말 것'이라는 지침을 위반했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "레퍼런스 이미지의 카메라 구도를 똑같이 차용해 프레이밍 지침을 어겼고, 인물의 표정이 강조되지 않았습니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S112sh6_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 샷에서 남자가 서 있던 보트의 금속 발판이 비어 있으며, 남자의 팔만 화면 왼쪽 밖에서 부자연스럽게 뻗어 나와 있습니다.",
     "fix_en": "Render the man's torso and legs standing on the metal boarding platform so his body properly connects to the outstretched arm.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Render the man's torso and legs standing on the metal boarding platform so his body properly connects to the outstretched arm.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S112sh6"
  }
 },
 "S112sh16": {
  "input_fingerprint": "57d145c858790280",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day, clear sky.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 정인우의 쓴웃음에 미소로 화답하며 따뜻한 눈빛을 보내는 얼굴 클로즈업.\n\nLOCATION (lock): On the sunlit island pier beside the departing fishing boat. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the bright harbor, boat, pier, clear sky, and the pair facing one another. Exclude the earlier offered-hand tension and focus on the woman's warm answering smile.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains on Geumwoldo without the destroyed pendant, while Inwoo remains aboard Hwang’s fishing boat.\n\nNo people appear unless the shot text itself says so.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "요구된 '얼굴 클로즈업'에 가장 근접하며, 이전 샷에서 확립된 두 사람이 마주보는 자세와 위치를 잘 유지했습니다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "클로즈업이 아닌 미디엄 샷으로 렌더링되었으며, 남자가 서로 마주보지 않고 뒤돌아 있어 연속성 지시를 위반했습니다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "근경의 남자 외에도 배경의 배 위에 지시되지 않은 추가 인물이 생성되어 심각한 위반(Hard violation)에 해당합니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S112sh8_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "카메라가 남자 어깨 너머로 여자를 바라보는 앵글이나, 여자 뒤로 바다와 배가 배경으로 나타나 공간적 연속성이 무너졌습니다. 레퍼런스에 따르면 남자는 배 위에, 여자는 방파제에서 배를 바라보고 있어야 하므로, 이 구도에서는 여자 배경에 방파제나 육지가 보여야 합니다.",
     "fix_en": "The spatial relationship is inverted, placing the water and boat behind the woman instead of the pier; this requires relocating the camera and rebuilding the scene's geometry to keep the man on the boat.",
     "needs_regeneration": false,
     "unfixable": true
    },
    {
     "issue_ko": "화면 왼쪽 전경에 보이는 남자의 상의가 밝은 회색으로 보입니다. 레퍼런스의 어둡고 때 탄 청회색 작업복과 일치하지 않습니다.",
     "fix_en": "Darken the man's visible clothing in the foreground left and add heavy dirt/wear textures to match the dark blue-grey coveralls from the reference.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Darken the man's visible clothing in the foreground left and add heavy dirt/wear textures to match the dark blue-grey coveralls from the reference.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S112sh8"
  }
 },
 "S113sh3": {
  "input_fingerprint": "a3420caea9779720",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 금월도 선착장에 홀로 선 채 멀어지는 배를 향해 빙그레 옅은 미소를 띠고 있는 김지후(강수리영)의 모습.\n\nLOCATION (lock): Alone on the island pier, facing the fishing boat as it pulls away across the water. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains alone on the Geumwoldo pier without her destroyed pendant as Inwoo’s boat departs.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (한국인 20대 초반 여성, 예쁘장한 타원형 얼굴, 긴 검은 생머리, 젊고 창백한 인상) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 3,
   "C": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "프롬프트가 요구한 콘크리트 선착장, 인물의 외양, 옅은 미소, 멀어지는 배의 구도를 모두 훌륭하게 구현했습니다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "화면에 프롬프트에서 금지한 텍스트(자막)가 포함되었으며, 선착장의 재질도 콘크리트가 아닌 나무로 잘못 묘사되었습니다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "선착장과 인물의 묘사는 좋으나, 배가 멀어지지 않고 카메라 쪽으로 다가오는 방향으로 그려져 지시사항과 어긋납니다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S112sh16_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:736787>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지후가 멀어지는 배를 향해 있어야 하나, 현재 시선 방향(화면 왼쪽)과 배의 위치(화면 오른쪽 배경)가 일치하지 않습니다.",
     "fix_en": "Move the departing blue fishing boat from the right background to the left side of the horizon so that it aligns with her gaze.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Move the departing blue fishing boat from the right background to the left side of the horizon so that it aligns with her gaze.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S112sh16"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "groupbg::낚시배_갑판": {
  "input_fingerprint": "78361b1d292ab8f0",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "낚시배_갑판",
    "tags": [
     "S101sh1",
     "S101sh6",
     "S101sh9",
     "S102sh12",
     "S113sh4",
     "S17sh1",
     "S17sh6",
     "S19sh2",
     "S19sh7",
     "S54sh3"
    ]
   },
   "context_sig": "fc68fbc5ec3e7898"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On the open deck of the fishing boat after it has left the island, facing toward the mainland.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n경찰서 형사과 사무실: 책상들이 배열되어 있고 서류와 모니터들이 복잡하게 놓인 형사과 사무실. 백색 형광등이 실내를 밝힌다. (특징: 파티션으로 나뉜 책상들; 컴퓨터 모니터; 널려 있는 서류 더미; 밝은 백색 형광등 조명) / 구형 산타페 SUV 내부: 오래된 구형 SUV 차량의 실내 공간. 계기판 불빛이 약하게 실내를 비춘다. (특징: 운전대; 낡은 가죽 시트; 계기판 조명; 차량 창문 너머로 보이는 새벽 풍경) / 금월도 숲속 만월제 도당: 숲 한가운데 둥글게 트인 공터. 거대한 바위와 제단이 있고 커다란 모닥불들이 타오른다. (특징: 성인 키 서너 배 높이의 거대한 암석 바위; 피 손자국으로 범벅이 된 바위 표면; 바위 앞 제단; 제단을 둥글게 감싸며 타오르는 커다란 모닥불들; 하늘 위 거대한 핏빛 보름달) / 해안마을 경찰 지구대 내부: 책상과 서류장, 오래된 소파가 단출하게 배치된 낡은 시골 지구대 사무실. (특징: 낡고 오래된 사무용 책상; 철제 관물대와 서류장; 유선 전화기; 오래된 가죽 소파) / 해안마을 선착장: 육지 쪽 항구의 선착장. 어선들이 정박해 있고 콘크리트 바닥에 폴리스 라인이 쳐져 있다. (특징: 콘크리트 항구 부두 바닥; 드럼통 모닥불; 시체 주변의 노란색 폴리스 라인; 구경꾼 무리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 달리는 낚시 어선 - 낮\n- 화면 가득 푸르게 넘실대는 바다에 5톤 짜리 낚시 어선이 하얀 포말을 만들며 지나간다.\n- 실외. 달리는 낚시 어선 - 낮\n- 실외. 황 씨의 낚시 어선 - 낮\n- 실외. 낚시 어선 - 낮\n- 실외. 낚시 어선 - 낮\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated: On the open deck of the fishing boat after it has left the island, facing toward the mainland.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n경찰서 형사과 사무실: 책상들이 배열되어 있고 서류와 모니터들이 복잡하게 놓인 형사과 사무실. 백색 형광등이 실내를 밝힌다. (특징: 파티션으로 나뉜 책상들; 컴퓨터 모니터; 널려 있는 서류 더미; 밝은 백색 형광등 조명) / 구형 산타페 SUV 내부: 오래된 구형 SUV 차량의 실내 공간. 계기판 불빛이 약하게 실내를 비춘다. (특징: 운전대; 낡은 가죽 시트; 계기판 조명; 차량 창문 너머로 보이는 새벽 풍경) / 금월도 숲속 만월제 도당: 숲 한가운데 둥글게 트인 공터. 거대한 바위와 제단이 있고 커다란 모닥불들이 타오른다. (특징: 성인 키 서너 배 높이의 거대한 암석 바위; 피 손자국으로 범벅이 된 바위 표면; 바위 앞 제단; 제단을 둥글게 감싸며 타오르는 커다란 모닥불들; 하늘 위 거대한 핏빛 보름달) / 해안마을 경찰 지구대 내부: 책상과 서류장, 오래된 소파가 단출하게 배치된 낡은 시골 지구대 사무실. (특징: 낡고 오래된 사무용 책상; 철제 관물대와 서류장; 유선 전화기; 오래된 가죽 소파) / 해안마을 선착장: 육지 쪽 항구의 선착장. 어선들이 정박해 있고 콘크리트 바닥에 폴리스 라인이 쳐져 있다. (특징: 콘크리트 항구 부두 바닥; 드럼통 모닥불; 시체 주변의 노란색 폴리스 라인; 구경꾼 무리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 달리는 낚시 어선 - 낮\n- 화면 가득 푸르게 넘실대는 바다에 5톤 짜리 낚시 어선이 하얀 포말을 만들며 지나간다.\n- 실외. 달리는 낚시 어선 - 낮\n- 실외. 황 씨의 낚시 어선 - 낮\n- 실외. 낚시 어선 - 낮\n- 실외. 낚시 어선 - 낮\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_낚시배_갑판_480abc.png",
  "asset_id": "e0250d01-9c2c-4d36-8993-1a8c3b047efb",
  "input_asset_ids": [
   "182068a3-22e7-4041-abf3-ea94efc83aeb"
  ],
  "origin_tag": "S113sh4",
  "place_text": "On the open deck of the fishing boat after it has left the island, facing toward the mainland.",
  "origin_inputs": {
   "place_text": "On the open deck of the fishing boat after it has left the island, facing toward the mainland.",
   "time_of_day_en": "day",
   "conti_asset_id": "182068a3-22e7-4041-abf3-ea94efc83aeb"
  }
 },
 "S113sh4::bgfirst_bg": {
  "input_fingerprint": "7ebceaf2afe622b6",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 바닷바람에 머리카락이 흩날리는 가운데 육지 쪽을 향해 시선을 고정하고 눈가에 맑은 눈물이 맺혀 있는 정인우의 얼굴 클로즈업.\n\nLOCATION (lock): On the open deck of the fishing boat after it has left the island, facing toward the mainland.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: On 정인우's landward side at eye level, dolly from close distance into an intimate facial close-up, ending with his face occupying roughly the central two-thirds of the frame. Keep him in a three-quarter turn rather than frontal portraiture, with his tear-bright eyes fixed beyond the frame and his wind-moved hair crossing the clean outline of his face.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: sea (sparkling in daylight) — The defocused water surface catches daylight behind his landward profile; used as Forms the soft background beyond 정인우 and preserves the boat's coastal setting during the facial push-in; fishing boat deck (moving toward the mainland) — Only the landward-facing portion behind 정인우 remains visible at the lower edge; used as Provides a minimal lower-frame reference that keeps the close-up physically anchored aboard the vessel.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 바닷바람에 머리카락이 흩날리는 가운데 육지 쪽을 향해 시선을 고정하고 눈가에 맑은 눈물이 맺혀 있는 정인우의 얼굴 클로즈업.\n\nLOCATION (lock): On the open deck of the fishing boat after it has left the island, facing toward the mainland.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: On 정인우's landward side at eye level, dolly from close distance into an intimate facial close-up, ending with his face occupying roughly the central two-thirds of the frame. Keep him in a three-quarter turn rather than frontal portraiture, with his tear-bright eyes fixed beyond the frame and his wind-moved hair crossing the clean outline of his face.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: sea (sparkling in daylight) — The defocused water surface catches daylight behind his landward profile; used as Forms the soft background beyond 정인우 and preserves the boat's coastal setting during the facial push-in; fishing boat deck (moving toward the mainland) — Only the landward-facing portion behind 정인우 remains visible at the lower edge; used as Provides a minimal lower-frame reference that keeps the close-up physically anchored aboard the vessel.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S113sh4__bgfirst_bg.png",
  "asset_id": "60fb45ef-6b41-4bb2-9717-afa25aa38981",
  "input_asset_ids": [
   "182068a3-22e7-4041-abf3-ea94efc83aeb",
   "e0250d01-9c2c-4d36-8993-1a8c3b047efb"
  ]
 },
 "S113sh4": {
  "input_fingerprint": "9d517c63313d08e5",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 바닷바람에 머리카락이 흩날리는 가운데 육지 쪽을 향해 시선을 고정하고 눈가에 맑은 눈물이 맺혀 있는 정인우의 얼굴 클로즈업.\n\nLOCATION (lock): On the open deck of the fishing boat after it has left the island, facing toward the mainland. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo remains aboard the departing fishing boat while Jihu stays behind on Geumwoldo.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 바닷바람에 머리카락이 흩날리는 가운데 육지 쪽을 향해 시선을 고정하고 눈가에 맑은 눈물이 맺혀 있는 정인우의 얼굴 클로즈업.\n\nLOCATION (lock): On the open deck of the fishing boat after it has left the island, facing toward the mainland. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo remains aboard the departing fishing boat while Jihu stays behind on Geumwoldo.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, with historical flashbacks to 1670 Joseon and the Japanese military era in Korea; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 바닷바람에 머리카락이 흩날리는 가운데 육지 쪽을 향해 시선을 고정하고 눈가에 맑은 눈물이 맺혀 있는 정인우의 얼굴 클로즈업.\n\nLOCATION (lock): On the open deck of the fishing boat after it has left the island, facing toward the mainland. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo remains aboard the departing fishing boat while Jihu stays behind on Geumwoldo.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (한국인 20대 중반 남성, 훤칠한 얼굴, 짧은 검은 머리, 단정한 이목구비) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S113sh4__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S113sh4.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:903809>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_낚시배_갑판_480abc.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:903809>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 얼굴 클로즈업 앵글과 로케이션 레퍼런스의 구조물(파란 선체, 조명 기둥)을 가장 정확히 재현했으나 의상이 다소 다릅니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "캐릭터 의상은 레퍼런스와 일치하나, 로케이션에 없는 흰색 선실 구조물이 추가되었고 샷 사이즈가 다소 넓습니다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "지정된 얼굴 클로즈업 앵글과 로케이션 레퍼런스의 구조물(파란 선체, 조명 기둥)을 가장 정확히 재현했으나 의상이 다소 다릅니다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "캐릭터 의상은 레퍼런스와 일치하나, 로케이션에 없는 흰색 선실 구조물이 추가되었고 샷 사이즈가 다소 넓습니다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "클로즈업 앵글, 눈물 맺힌 표정을 잘 구현했으며 특히 레퍼런스의 복장(조끼와 재킷)을 매우 정확하게 반영했습니다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "바람에 흩날리는 머리카락과 배경의 재현도는 우수하나, 캐릭터 레퍼런스의 핵심 복장을 완전히 누락하여 우선순위에서 밀립니다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "클로즈업 앵글, 눈물 맺힌 표정을 잘 구현했으며 특히 레퍼런스의 복장(조끼와 재킷)을 매우 정확하게 반영했습니다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "바람에 흩날리는 머리카락과 배경의 재현도는 우수하나, 캐릭터 레퍼런스의 핵심 복장을 완전히 누락하여 우선순위에서 밀립니다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 12
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 11,
   "B": 12
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "지정된 얼굴 클로즈업 앵글과 로케이션 레퍼런스의 구조물(파란 선체, 조명 기둥)을 가장 정확히 재현했으나 의상이 다소 다릅니다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "캐릭터 의상은 레퍼런스와 일치하나, 로케이션에 없는 흰색 선실 구조물이 추가되었고 샷 사이즈가 다소 넓습니다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_낚시배_갑판_480abc.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903809>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "기준 장소 사진에 의하면 우측 배경은 파란색 선체 벽과 은색 난간만 있어야 하나, 사진에 없는 밧줄이 감긴 기둥이 임의로 생성되어 장소의 구조적 진실성을 위반함.",
     "fix_en": "Remove the vertical post and coiled ropes on the right side of the background, replacing them with the continuous blue bulwark, silver railing, and sea.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the vertical post and coiled ropes on the right side of the background, replacing them with the continuous blue bulwark, silver railing, and sea.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S113sh4__bgfirst_bg.png",
   "bg_asset_id": "60fb45ef-6b41-4bb2-9717-afa25aa38981",
   "bg_record_key": "S113sh4::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "낚시배_갑판",
   "groupbg_asset_id": "e0250d01-9c2c-4d36-8993-1a8c3b047efb"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S2sh1::bgfirst_bg": {
  "input_fingerprint": "996d2976df071029",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 얼기설기 지어진 나무 창고 안에서 과거의 조선인 육혈귀 생존자들이 서로 부둥켜안고 두려움에 몸을 움츠린 상태.\n\nLOCATION (lock): Inside the roughly patched wooden salt-storehouse, away from the entrance and lit by daytime leaking through its gaps.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height near the rear interior, hold a wide static composition with the clustered survivors in the middle foreground and the door aligned beyond them. The loosely assembled walls and supports close around their hunched bodies, while the forward axis leaves the door as the inescapable destination of the next move.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: huddled Joseon survivors in the middle-center of the frame, midground; warehouse door in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 판자와 각목으로 덧댄 창고 내부 (얼기설기 덧대어 만든 상태) — The inner faces of the patched walls and exposed supports surround the survivors; used as Compresses the group inside an insecure, improvised enclosure; 창고 문 (이 순간 닫혀 있는 상태) — The interior-facing side is visible beyond the clustered survivors; used as Sits on the forward camera axis as the source of approaching danger.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: High-contrast monochrome daylight appropriate to the interior holds the survivors in deep tonal separation from the enclosing structure.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 얼기설기 지어진 나무 창고 안에서 과거의 조선인 육혈귀 생존자들이 서로 부둥켜안고 두려움에 몸을 움츠린 상태.\n\nLOCATION (lock): Inside the roughly patched wooden salt-storehouse, away from the entrance and lit by daytime leaking through its gaps.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height near the rear interior, hold a wide static composition with the clustered survivors in the middle foreground and the door aligned beyond them. The loosely assembled walls and supports close around their hunched bodies, while the forward axis leaves the door as the inescapable destination of the next move.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: huddled Joseon survivors in the middle-center of the frame, midground; warehouse door in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 판자와 각목으로 덧댄 창고 내부 (얼기설기 덧대어 만든 상태) — The inner faces of the patched walls and exposed supports surround the survivors; used as Compresses the group inside an insecure, improvised enclosure; 창고 문 (이 순간 닫혀 있는 상태) — The interior-facing side is visible beyond the clustered survivors; used as Sits on the forward camera axis as the source of approaching danger.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: High-contrast monochrome daylight appropriate to the interior holds the survivors in deep tonal separation from the enclosing structure.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S2sh1__bgfirst_bg.png",
  "asset_id": "9086b5a8-6352-4d3f-a0f6-bd12963d286a",
  "input_asset_ids": [
   "6dacf3a9-3929-45e0-aaf2-24e27bb11e81",
   "64d63671-c7fb-4465-a3a2-dd326e5b074b"
  ]
 },
 "S2sh1": {
  "input_fingerprint": "a38762bd23869262",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 얼기설기 지어진 나무 창고 안에서 과거의 조선인 육혈귀 생존자들이 서로 부둥켜안고 두려움에 몸을 움츠린 상태.\n\nLOCATION (lock): Inside the roughly patched wooden salt-storehouse, away from the entrance and lit by daytime leaking through its gaps. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height near the rear interior, hold a wide static composition with the clustered survivors in the middle foreground and the door aligned beyond them. The loosely assembled walls and supports close around their hunched bodies, while the forward axis leaves the door as the inescapable destination of the next move.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: huddled Joseon survivors in the middle-center of the frame, midground; warehouse door in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 판자와 각목으로 덧댄 창고 내부 (얼기설기 덧대어 만든 상태) — The inner faces of the patched walls and exposed supports surround the survivors; used as Compresses the group inside an insecure, improvised enclosure; 창고 문 (이 순간 닫혀 있는 상태) — The interior-facing side is visible beyond the clustered survivors; used as Sits on the forward camera axis as the source of approaching danger.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: High-contrast monochrome daylight appropriate to the interior holds the survivors in deep tonal separation from the enclosing structure.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 과거의 조선인 육혈귀 생존자 (Korean adult, dark hair, gaunt weathered face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 얼기설기 지어진 나무 창고 안에서 과거의 조선인 육혈귀 생존자들이 서로 부둥켜안고 두려움에 몸을 움츠린 상태.\n\nLOCATION (lock): Inside the roughly patched wooden salt-storehouse, away from the entrance and lit by daytime leaking through its gaps. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height near the rear interior, hold a wide static composition with the clustered survivors in the middle foreground and the door aligned beyond them. The loosely assembled walls and supports close around their hunched bodies, while the forward axis leaves the door as the inescapable destination of the next move.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: huddled Joseon survivors in the middle-center of the frame, midground; warehouse door in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 판자와 각목으로 덧댄 창고 내부 (얼기설기 덧대어 만든 상태) — The inner faces of the patched walls and exposed supports surround the survivors; used as Compresses the group inside an insecure, improvised enclosure; 창고 문 (이 순간 닫혀 있는 상태) — The interior-facing side is visible beyond the clustered survivors; used as Sits on the forward camera axis as the source of approaching danger.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: High-contrast monochrome daylight appropriate to the interior holds the survivors in deep tonal separation from the enclosing structure.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 과거의 조선인 육혈귀 생존자 (Korean adult, dark hair, gaunt weathered face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 얼기설기 지어진 나무 창고 안에서 과거의 조선인 육혈귀 생존자들이 서로 부둥켜안고 두려움에 몸을 움츠린 상태.\n\nLOCATION (lock): Inside the roughly patched wooden salt-storehouse, away from the entrance and lit by daytime leaking through its gaps. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height near the rear interior, hold a wide static composition with the clustered survivors in the middle foreground and the door aligned beyond them. The loosely assembled walls and supports close around their hunched bodies, while the forward axis leaves the door as the inescapable destination of the next move.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: huddled Joseon survivors in the middle-center of the frame, midground; warehouse door in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 판자와 각목으로 덧댄 창고 내부 (얼기설기 덧대어 만든 상태) — The inner faces of the patched walls and exposed supports surround the survivors; used as Compresses the group inside an insecure, improvised enclosure; 창고 문 (이 순간 닫혀 있는 상태) — The interior-facing side is visible beyond the clustered survivors; used as Sits on the forward camera axis as the source of approaching danger.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: High-contrast monochrome daylight appropriate to the interior holds the survivors in deep tonal separation from the enclosing structure.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 과거의 조선인 육혈귀 생존자 (Korean adult, dark hair, gaunt weathered face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S2sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S2sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 과거의 조선인 육혈귀 생존자: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:934378>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L02B02.png"
    },
    {
     "label": "CHARACTER REFERENCE — 과거의 조선인 육혈귀 생존자: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:934378>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 후방 실내 가슴 높이에서 전방 축을 따라 창고 문(화면 중앙 배경, 닫힘)을 바라본다. 문은 프레임 중앙 상단 배경에 정확히 정렬. 생존자들은 대부분 시선을 아래로 떨구거나 서로의 어깨에 얼굴을 묻고 있어 문 쪽을 향한 경계 시선은 약하다. 인물 무리는 중앙 전경~중경에 위치.",
      "built_space": "레퍼런스 창고와 구조 일치: 좌측 벽의 긴 수평 각목 보, 우측 벽의 빗장 달린 판자문 흔적, 정면 판자벽과 틈새 빛, 노출된 서까래. 다만 프레임 중앙 배경의 '문'은 레퍼런스의 정면 벽 위치에 새로 만들어진 문으로, 우측 벽에도 또 하나의 문(경첩·빗장)이 보여 문이 둘로 읽힌다. 바닥은 흙바닥으로 일치.",
      "entities": "성인 조선인 남녀 15명 이상, 헝클어진 낡은 무명 저고리·치마, 수척한 얼굴. 다만 캐릭터 레퍼런스의 백발 검은 한복 노파는 명확히 식별되지 않는다. 인원이 과다하여 '생존자 무리'라기보다 군집으로 보이며, 뒤쪽 인물 일부는 얼굴이 뭉개져 있다.",
      "hard_violations": [],
      "physics": "모든 인물이 흙바닥에 무릎을 꿇거나 주저앉아 있고 서로의 어깨·등을 손으로 감싸 접촉 지지가 명확하다. 공중에 뜬 신체 없음. 다만 중앙 뒤쪽 몇몇 인물은 몸통이 겹쳐 팔의 소속이 모호한 부분이 있다."
     },
     {
      "label": "B",
      "direction": "카메라는 후방 실내 가슴 높이, 정면 전방축. 닫힌 판자문이 화면 중앙 배경에 정확히 정렬되고 그 앞 중경 중앙에 다섯 명이 뭉쳐 있다. 중앙 노파와 우측 남자는 카메라(=문 방향 축)를 향해 시선을 두어 다가오는 위협 축을 바라보는 연기가 읽힌다. 나머지는 서로에게 얼굴을 붙이고 있다.",
      "built_space": "레퍼런스와 구조 일치: 좌측 벽 수평 각목 보, 우측 벽 빗장·경첩 달린 판자문, 정면 판자벽 틈새 빛, 노출 서까래, 흙바닥. 정면 벽 중앙에 닫힌 문면이 배치되어 전방 축의 목적지로 기능. 벽 두께·판자 이음새·빛샘 라인이 자연스럽고 중복 구조물 없음.",
      "entities": "성인 조선인 5인, 검게 그을린 수척한 얼굴, 헝클어진 머리, 해진 무명옷. 중앙 인물은 캐릭터 레퍼런스의 백발 쪽진 머리·검은 한복 노파와 얼굴·머리·의상이 뚜렷이 일치. 프롬프트가 금지한 외부 인물 없음, 눈은 정상 인간 홍채/공막.",
      "hard_violations": [],
      "physics": "다섯 명 모두 흙바닥에 무릎을 꿇고 앉아 있고, 서로의 팔이 몸통을 감싸 안아 접촉 지지가 명확하다. 옷 주름이 무게에 따라 늘어지고 무릎 아래 천이 바닥에 눌린다. 공중 부양이나 지지 없는 신체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "와이드 구도와 로케이션 일치는 양호하나, 인물 수가 과도해 '생존자들이 부둥켜안은' 밀집 무리라기보다 군중으로 읽히고 캐릭터 레퍼런스의 노파가 식별되지 않으며, 정면 문과 우측 문이 동시에 보여 전방 축의 목적지가 분산된다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "가슴 높이 후방 와이드, 중앙 중경의 웅크린 생존자 무리와 그 너머 정중앙에 정렬된 닫힌 문까지 프레임 레이아웃을 정확히 실현했고, 캐릭터 레퍼런스 노파의 얼굴·백발·검은 한복이 일치하며 지지·고대비 흑백 톤도 규정대로다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "와이드 구도와 로케이션 일치는 양호하나, 인물 수가 과도해 '생존자들이 부둥켜안은' 밀집 무리라기보다 군중으로 읽히고 캐릭터 레퍼런스의 노파가 식별되지 않으며, 정면 문과 우측 문이 동시에 보여 전방 축의 목적지가 분산된다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "가슴 높이 후방 와이드, 중앙 중경의 웅크린 생존자 무리와 그 너머 정중앙에 정렬된 닫힌 문까지 프레임 레이아웃을 정확히 실현했고, 캐릭터 레퍼런스 노파의 얼굴·백발·검은 한복이 일치하며 지지·고대비 흑백 톤도 규정대로다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 후방 실내 가슴 높이에서 전방 축을 따라 창고 문(화면 중앙 배경, 닫힘)을 바라본다. 문은 프레임 중앙 상단 배경에 정확히 정렬. 생존자들은 대부분 시선을 아래로 떨구거나 서로의 어깨에 얼굴을 묻고 있어 문 쪽을 향한 경계 시선은 약하다. 인물 무리는 중앙 전경~중경에 위치.",
      "built_space": "레퍼런스 창고와 구조 일치: 좌측 벽의 긴 수평 각목 보, 우측 벽의 빗장 달린 판자문 흔적, 정면 판자벽과 틈새 빛, 노출된 서까래. 다만 프레임 중앙 배경의 '문'은 레퍼런스의 정면 벽 위치에 새로 만들어진 문으로, 우측 벽에도 또 하나의 문(경첩·빗장)이 보여 문이 둘로 읽힌다. 바닥은 흙바닥으로 일치.",
      "entities": "성인 조선인 남녀 15명 이상, 헝클어진 낡은 무명 저고리·치마, 수척한 얼굴. 다만 캐릭터 레퍼런스의 백발 검은 한복 노파는 명확히 식별되지 않는다. 인원이 과다하여 '생존자 무리'라기보다 군집으로 보이며, 뒤쪽 인물 일부는 얼굴이 뭉개져 있다.",
      "hard_violations": [],
      "physics": "모든 인물이 흙바닥에 무릎을 꿇거나 주저앉아 있고 서로의 어깨·등을 손으로 감싸 접촉 지지가 명확하다. 공중에 뜬 신체 없음. 다만 중앙 뒤쪽 몇몇 인물은 몸통이 겹쳐 팔의 소속이 모호한 부분이 있다."
     },
     {
      "label": "B",
      "direction": "카메라는 후방 실내 가슴 높이, 정면 전방축. 닫힌 판자문이 화면 중앙 배경에 정확히 정렬되고 그 앞 중경 중앙에 다섯 명이 뭉쳐 있다. 중앙 노파와 우측 남자는 카메라(=문 방향 축)를 향해 시선을 두어 다가오는 위협 축을 바라보는 연기가 읽힌다. 나머지는 서로에게 얼굴을 붙이고 있다.",
      "built_space": "레퍼런스와 구조 일치: 좌측 벽 수평 각목 보, 우측 벽 빗장·경첩 달린 판자문, 정면 판자벽 틈새 빛, 노출 서까래, 흙바닥. 정면 벽 중앙에 닫힌 문면이 배치되어 전방 축의 목적지로 기능. 벽 두께·판자 이음새·빛샘 라인이 자연스럽고 중복 구조물 없음.",
      "entities": "성인 조선인 5인, 검게 그을린 수척한 얼굴, 헝클어진 머리, 해진 무명옷. 중앙 인물은 캐릭터 레퍼런스의 백발 쪽진 머리·검은 한복 노파와 얼굴·머리·의상이 뚜렷이 일치. 프롬프트가 금지한 외부 인물 없음, 눈은 정상 인간 홍채/공막.",
      "hard_violations": [],
      "physics": "다섯 명 모두 흙바닥에 무릎을 꿇고 앉아 있고, 서로의 팔이 몸통을 감싸 안아 접촉 지지가 명확하다. 옷 주름이 무게에 따라 늘어지고 무릎 아래 천이 바닥에 눌린다. 공중 부양이나 지지 없는 신체 없음."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 창고 후면 가슴 높이에서 전방 축을 따라 놓여 있고, 중앙 중경에 5명의 생존자 무리가 서로 끌어안고 앉아 있다. 그 너머 정면 중앙에 판자로 된 닫힌 창고 벽/문면이 정렬되어 전진 축의 종착점으로 읽힌다. 중앙 노파의 시선은 렌즈 방향(전방 문 쪽 축)을 향하고, 좌우 인물들의 시선은 각각 아래·전방으로 향하며 서로에게 매달려 있다. 무기·지시물은 없다.",
      "built_space": "레퍼런스 창고 구조와 일치: 좌측 벽의 긴 수평 각목 두 줄, 우측 벽의 빗장·경첩이 달린 판자문, 상부 서까래, 널판 사이로 새어 드는 주광 슬릿. 정면 안쪽 벽은 세로 판자와 중간 가로 보강재로 덧대어져 있으며 문/출입 면이 중앙에 정렬. 벽·기둥 수가 중복되거나 여분의 구조물이 생성된 곳은 없다. 다만 레퍼런스의 문이 우측 벽에 있는 반면 여기서는 정면에도 문 유사 면이 있는데, 이는 프롬프트가 요구한 '전방 축의 문' 배치와 부합.",
      "entities": "인물 5명 전원 한국인, 검게 그을리고 야윈 얼굴, 흐트러진 검은 머리, 낡고 해진 흰 무명 저고리·치마. 중앙 인물은 캐릭터 레퍼런스의 백발·검정 한복 노파와 얼굴·머리·체구가 명확히 일치. 눈은 정상 홍채·공막의 해부학적 인간 눈. 텍스트·워터마크 없음. 고대비 흑백 주광 유지.",
      "hard_violations": [],
      "physics": "5인 모두 흙바닥에 무릎 꿇거나 주저앉아 지면이 체중을 받치고 있다. 각자의 팔이 옆 사람의 어깨·팔뚝을 실제로 감싸 접촉점이 명확하고, 옷 주름이 눌린 방향과 일치. 공중에 뜬 신체나 지지 없는 물체 없음. 무릎·정강이·발이 바닥 먼지와 닿아 그림자를 만든다."
     },
     {
      "label": "B",
      "direction": "카메라는 후면 가슴~약간 높은 위치에서 전방을 향하고, 정면 중앙 배경에 닫힌 판자문이 정렬되어 있다. 무리는 프레임 중앙보다 아래·앞쪽으로 크게 치우쳐 전경을 채우며, 전열 인물 다수가 카메라에 등을 보여 얼굴·시선이 보이지 않는다. 시선은 대부분 아래 또는 서로에게 향하고 문 쪽 축을 인지하는 인물은 거의 없다.",
      "built_space": "레퍼런스 구조와 대체로 일치: 좌측 수평 각목, 우측 벽 판자문과 경첩, 벽 틈 광선. 정면 중앙 문도 존재. 구조 부재 중복은 없으나 좌측 상단 각목이 벽 바깥으로 어색하게 튀어나오고 정면 문 주변 벽 패널 이음이 다소 뭉개져 있다.",
      "entities": "인물이 약 14명 이상으로 대량 증식되어 있으며 다수가 유사한 흰 저고리·땋은 머리의 젊은 여성으로 반복된다. 캐릭터 레퍼런스의 백발 노파는 프레임 안에서 확실히 식별되지 않아(우측 중경 인물이 유사하나 백발·검정 한복이 아님) 지정 인물 일치가 불명. 뒷줄 일부 인물은 몸통·팔이 다른 인물과 융합되어 팔 개수와 어깨 연결이 불명확하다. 텍스트·워터마크 없음.",
      "hard_violations": [
       "뒷줄 중앙~좌측에서 인물들의 몸통·팔·머리가 서로 융합되어 소속 불명의 여분 팔과 반쯤 그려진 신체가 생성됨(해부학적으로 불가능한 신체 중첩)"
      ],
      "physics": "대부분 바닥에 앉아 지면 지지를 받고 있어 공중 부양은 없다. 다만 중경 뒷줄 몇몇 인물은 하반신이 앞사람에 완전히 가려진 채 어깨만 떠 있는 듯 배치되어 지지 근거가 불분명하며, 어깨를 감싼 팔 중 일부는 주인을 특정할 수 없다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "가슴 높이 후면 와이드, 중앙 중경의 껴안은 생존자 무리와 그 너머 정렬된 닫힌 문까지 프레임 지시를 정확히 충족하고, 캐릭터 레퍼런스 노파가 중앙에 명확히 일치하며 지지·해부학 문제도 없다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "장소와 문 정렬은 맞으나 인물이 14명 이상으로 증식되고 뒷줄에서 신체·팔이 융합되는 해부학적 불가능이 발생했으며, 무리가 전경을 과하게 채우고 지정 캐릭터(백발 노파)가 식별되지 않는다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "가슴 높이 후면 와이드, 중앙 중경의 껴안은 생존자 무리와 그 너머 정렬된 닫힌 문까지 프레임 지시를 정확히 충족하고, 캐릭터 레퍼런스 노파가 중앙에 명확히 일치하며 지지·해부학 문제도 없다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "장소와 문 정렬은 맞으나 인물이 14명 이상으로 증식되고 뒷줄에서 신체·팔이 융합되는 해부학적 불가능이 발생했으며, 무리가 전경을 과하게 채우고 지정 캐릭터(백발 노파)가 식별되지 않는다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "카메라는 창고 후면 가슴 높이에서 전방 축을 따라 놓여 있고, 중앙 중경에 5명의 생존자 무리가 서로 끌어안고 앉아 있다. 그 너머 정면 중앙에 판자로 된 닫힌 창고 벽/문면이 정렬되어 전진 축의 종착점으로 읽힌다. 중앙 노파의 시선은 렌즈 방향(전방 문 쪽 축)을 향하고, 좌우 인물들의 시선은 각각 아래·전방으로 향하며 서로에게 매달려 있다. 무기·지시물은 없다.",
      "built_space": "레퍼런스 창고 구조와 일치: 좌측 벽의 긴 수평 각목 두 줄, 우측 벽의 빗장·경첩이 달린 판자문, 상부 서까래, 널판 사이로 새어 드는 주광 슬릿. 정면 안쪽 벽은 세로 판자와 중간 가로 보강재로 덧대어져 있으며 문/출입 면이 중앙에 정렬. 벽·기둥 수가 중복되거나 여분의 구조물이 생성된 곳은 없다. 다만 레퍼런스의 문이 우측 벽에 있는 반면 여기서는 정면에도 문 유사 면이 있는데, 이는 프롬프트가 요구한 '전방 축의 문' 배치와 부합.",
      "entities": "인물 5명 전원 한국인, 검게 그을리고 야윈 얼굴, 흐트러진 검은 머리, 낡고 해진 흰 무명 저고리·치마. 중앙 인물은 캐릭터 레퍼런스의 백발·검정 한복 노파와 얼굴·머리·체구가 명확히 일치. 눈은 정상 홍채·공막의 해부학적 인간 눈. 텍스트·워터마크 없음. 고대비 흑백 주광 유지.",
      "hard_violations": [],
      "physics": "5인 모두 흙바닥에 무릎 꿇거나 주저앉아 지면이 체중을 받치고 있다. 각자의 팔이 옆 사람의 어깨·팔뚝을 실제로 감싸 접촉점이 명확하고, 옷 주름이 눌린 방향과 일치. 공중에 뜬 신체나 지지 없는 물체 없음. 무릎·정강이·발이 바닥 먼지와 닿아 그림자를 만든다."
     },
     {
      "label": "A",
      "direction": "카메라는 후면 가슴~약간 높은 위치에서 전방을 향하고, 정면 중앙 배경에 닫힌 판자문이 정렬되어 있다. 무리는 프레임 중앙보다 아래·앞쪽으로 크게 치우쳐 전경을 채우며, 전열 인물 다수가 카메라에 등을 보여 얼굴·시선이 보이지 않는다. 시선은 대부분 아래 또는 서로에게 향하고 문 쪽 축을 인지하는 인물은 거의 없다.",
      "built_space": "레퍼런스 구조와 대체로 일치: 좌측 수평 각목, 우측 벽 판자문과 경첩, 벽 틈 광선. 정면 중앙 문도 존재. 구조 부재 중복은 없으나 좌측 상단 각목이 벽 바깥으로 어색하게 튀어나오고 정면 문 주변 벽 패널 이음이 다소 뭉개져 있다.",
      "entities": "인물이 약 14명 이상으로 대량 증식되어 있으며 다수가 유사한 흰 저고리·땋은 머리의 젊은 여성으로 반복된다. 캐릭터 레퍼런스의 백발 노파는 프레임 안에서 확실히 식별되지 않아(우측 중경 인물이 유사하나 백발·검정 한복이 아님) 지정 인물 일치가 불명. 뒷줄 일부 인물은 몸통·팔이 다른 인물과 융합되어 팔 개수와 어깨 연결이 불명확하다. 텍스트·워터마크 없음.",
      "hard_violations": [
       "뒷줄 중앙~좌측에서 인물들의 몸통·팔·머리가 서로 융합되어 소속 불명의 여분 팔과 반쯤 그려진 신체가 생성됨(해부학적으로 불가능한 신체 중첩)"
      ],
      "physics": "대부분 바닥에 앉아 지면 지지를 받고 있어 공중 부양은 없다. 다만 중경 뒷줄 몇몇 인물은 하반신이 앞사람에 완전히 가려진 채 어깨만 떠 있는 듯 배치되어 지지 근거가 불분명하며, 어깨를 감싼 팔 중 일부는 주인을 특정할 수 없다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 10,
     "B": 17
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "readings": [
   {
    "label": "A",
    "direction": "카메라는 후방 실내 가슴 높이에서 전방 축을 따라 창고 문(화면 중앙 배경, 닫힘)을 바라본다. 문은 프레임 중앙 상단 배경에 정확히 정렬. 생존자들은 대부분 시선을 아래로 떨구거나 서로의 어깨에 얼굴을 묻고 있어 문 쪽을 향한 경계 시선은 약하다. 인물 무리는 중앙 전경~중경에 위치.",
    "built_space": "레퍼런스 창고와 구조 일치: 좌측 벽의 긴 수평 각목 보, 우측 벽의 빗장 달린 판자문 흔적, 정면 판자벽과 틈새 빛, 노출된 서까래. 다만 프레임 중앙 배경의 '문'은 레퍼런스의 정면 벽 위치에 새로 만들어진 문으로, 우측 벽에도 또 하나의 문(경첩·빗장)이 보여 문이 둘로 읽힌다. 바닥은 흙바닥으로 일치.",
    "entities": "성인 조선인 남녀 15명 이상, 헝클어진 낡은 무명 저고리·치마, 수척한 얼굴. 다만 캐릭터 레퍼런스의 백발 검은 한복 노파는 명확히 식별되지 않는다. 인원이 과다하여 '생존자 무리'라기보다 군집으로 보이며, 뒤쪽 인물 일부는 얼굴이 뭉개져 있다.",
    "hard_violations": [],
    "physics": "모든 인물이 흙바닥에 무릎을 꿇거나 주저앉아 있고 서로의 어깨·등을 손으로 감싸 접촉 지지가 명확하다. 공중에 뜬 신체 없음. 다만 중앙 뒤쪽 몇몇 인물은 몸통이 겹쳐 팔의 소속이 모호한 부분이 있다."
   },
   {
    "label": "B",
    "direction": "카메라는 후방 실내 가슴 높이, 정면 전방축. 닫힌 판자문이 화면 중앙 배경에 정확히 정렬되고 그 앞 중경 중앙에 다섯 명이 뭉쳐 있다. 중앙 노파와 우측 남자는 카메라(=문 방향 축)를 향해 시선을 두어 다가오는 위협 축을 바라보는 연기가 읽힌다. 나머지는 서로에게 얼굴을 붙이고 있다.",
    "built_space": "레퍼런스와 구조 일치: 좌측 벽 수평 각목 보, 우측 벽 빗장·경첩 달린 판자문, 정면 판자벽 틈새 빛, 노출 서까래, 흙바닥. 정면 벽 중앙에 닫힌 문면이 배치되어 전방 축의 목적지로 기능. 벽 두께·판자 이음새·빛샘 라인이 자연스럽고 중복 구조물 없음.",
    "entities": "성인 조선인 5인, 검게 그을린 수척한 얼굴, 헝클어진 머리, 해진 무명옷. 중앙 인물은 캐릭터 레퍼런스의 백발 쪽진 머리·검은 한복 노파와 얼굴·머리·의상이 뚜렷이 일치. 프롬프트가 금지한 외부 인물 없음, 눈은 정상 인간 홍채/공막.",
    "hard_violations": [],
    "physics": "다섯 명 모두 흙바닥에 무릎을 꿇고 앉아 있고, 서로의 팔이 몸통을 감싸 안아 접촉 지지가 명확하다. 옷 주름이 무게에 따라 늘어지고 무릎 아래 천이 바닥에 눌린다. 공중 부양이나 지지 없는 신체 없음."
   }
  ],
  "totals": {
   "A": 10,
   "B": 17
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "와이드 구도와 로케이션 일치는 양호하나, 인물 수가 과도해 '생존자들이 부둥켜안은' 밀집 무리라기보다 군중으로 읽히고 캐릭터 레퍼런스의 노파가 식별되지 않으며, 정면 문과 우측 문이 동시에 보여 전방 축의 목적지가 분산된다."
   },
   {
    "label": "B",
    "score": 9,
    "verdict_ko": "가슴 높이 후방 와이드, 중앙 중경의 웅크린 생존자 무리와 그 너머 정중앙에 정렬된 닫힌 문까지 프레임 레이아웃을 정확히 실현했고, 캐릭터 레퍼런스 노파의 얼굴·백발·검은 한복이 일치하며 지지·고대비 흑백 톤도 규정대로다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L02B02.png"
   },
   {
    "label": "CHARACTER REFERENCE — 과거의 조선인 육혈귀 생존자: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:934378>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "카메라 프레이밍 지시를 위반하여 창고 문이 화면 중앙 배경이 아닌 우측 벽에 위치함.",
     "fix_en": "Replace the wooden boards in the center of the background wall with a closed, roughly patched wooden warehouse door, and change the existing door on the right wall into plain vertical wooden wall boards, preserving the current camera framing, the survivors' poses, clothing, and lighting exactly as they are.",
     "needs_regeneration": true
    },
    {
     "issue_ko": "화면 우측 남성의 팔 아래, 할머니의 왼팔을 잡고 있는 주인이 없는 정체불명의 손이 있음.",
     "fix_en": "Remove the extra hand gripping the older woman's left arm just below the man's left arm, replacing it with the continuous black fabric of the older woman's sleeve, keeping all characters, their poses, clothing, and the background strictly unchanged.",
     "needs_regeneration": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the wooden boards in the center of the background wall with a closed, roughly patched wooden warehouse door, and change the existing door on the right wall into plain vertical wooden wall boards, preserving the current camera framing, the survivors' poses, clothing, and lighting exactly as they are.\n- Remove the extra hand gripping the older woman's left arm just below the man's left arm, replacing it with the continuous black fabric of the older woman's sleeve, keeping all characters, their poses, clothing, and the background strictly unchanged.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "요구된 후방 가슴높이 와이드, 중앙 중경의 부둥켜안은 생존자 무리, 배경 중앙의 문 축을 모두 충족하고, 레퍼런스의 틈새 광선·문 경첩·빗장 등 고정 설비와 고대비 흑백 낮빛을 가장 충실히 유지한다.  ★위반: [gpt] 요구된 중앙의 닫힌 창고 문 외에 오른쪽 벽에 두 번째 완성된 문을 추가하여 출입구를 중복시켰다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "정면 문의 빗장·걸쇠를 명확히 한 이득은 있으나, 계약이 요구하지 않은 손실로 우측 문의 경첩·빗장 디테일과 벽 틈으로 새어드는 낮빛 광선이 대거 사라져 '틈으로 스미는 주광' 조명 조건과 로케이션 고정 설비가 약화되고 중앙 팔 부분도 뭉개졌다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.667,
      "B": 1.75
     },
     "adjusted": {
      "A": 1.417,
      "B": 1.75
     },
     "violations": {
      "A": [
       "[gpt] 요구된 중앙의 닫힌 창고 문 외에 오른쪽 벽에 두 번째 완성된 문을 추가하여 출입구를 중복시켰다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "B"
     },
     "agreed": false
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "요구된 후방 가슴높이 와이드, 중앙 중경의 부둥켜안은 생존자 무리, 배경 중앙의 문 축을 모두 충족하고, 레퍼런스의 틈새 광선·문 경첩·빗장 등 고정 설비와 고대비 흑백 낮빛을 가장 충실히 유지한다.  ★위반: [gpt] 요구된 중앙의 닫힌 창고 문 외에 오른쪽 벽에 두 번째 완성된 문을 추가하여 출입구를 중복시켰다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "정면 문의 빗장·걸쇠를 명확히 한 이득은 있으나, 계약이 요구하지 않은 손실로 우측 문의 경첩·빗장 디테일과 벽 틈으로 새어드는 낮빛 광선이 대거 사라져 '틈으로 스미는 주광' 조명 조건과 로케이션 고정 설비가 약화되고 중앙 팔 부분도 뭉개졌다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "인물·구도는 B와 사실상 동일하면서, '무리 너머 정중앙 배경에 닫힌 창고 문'이라는 프레임 레이아웃 핵심을 빗장·경첩이 보이는 정면 문으로 명확히 충족하고 문이 하나뿐이라 공간 판독도 깔끔하다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "인물·조명은 동일하나 정중앙 배경의 문 식별 요소가 사라지고 우측 벽에 별도의 문이 강조되어, 계약이 지정한 '전방 축 중앙의 문' 방향성과 문 배치 판독이 약해진 것이 순손실이다.  ★위반: [gpt] 중앙 축의 닫힌 창고 문 외에 오른쪽 벽에 철물 달린 두 번째 문을 함께 만들어, 계약이 지정한 단일 문의 수와 배치를 중복시켰다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.321
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.071
     },
     "violations": {
      "B": [
       "[gpt] 중앙 축의 닫힌 창고 문 외에 오른쪽 벽에 철물 달린 두 번째 문을 함께 만들어, 계약이 지정한 단일 문의 수와 배치를 중복시켰다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "인물·구도는 B와 사실상 동일하면서, '무리 너머 정중앙 배경에 닫힌 창고 문'이라는 프레임 레이아웃 핵심을 빗장·경첩이 보이는 정면 문으로 명확히 충족하고 문이 하나뿐이라 공간 판독도 깔끔하다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "인물·조명은 동일하나 정중앙 배경의 문 식별 요소가 사라지고 우측 벽에 별도의 문이 강조되어, 계약이 지정한 '전방 축 중앙의 문' 방향성과 문 배치 판독이 약해진 것이 순손실이다.  ★위반: [gpt] 중앙 축의 닫힌 창고 문 외에 오른쪽 벽에 철물 달린 두 번째 문을 함께 만들어, 계약이 지정한 단일 문의 수와 배치를 중복시켰다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 14
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S2sh1__bgfirst_bg.png",
   "bg_asset_id": "9086b5a8-6352-4d3f-a0f6-bd12963d286a",
   "bg_record_key": "S2sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S3sh11": {
  "input_fingerprint": "3e633fe50690d9b6",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day transitioning to night, then morning.\n\nSHOT TEXT (authoritative, Korean): 그분(검은 연기 형체)의 길쭉한 검은 손이 조선인 촌주의 목을 강하게 틀어쥔 순간의 클로즈업.\n\nLOCATION (lock): The open ground immediately in front of the old wooden salt-storehouse, amid the reeds and fallen soldiers. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At neck height beside both figures, isolate the elongated black hand clamped around the village chief's throat in a tight close frame. The chief's pulled neck and the figure's extended wrist cross the center diagonally, while most of both faces remain beyond the upper frame edge.\n- FRAMING SCALE: close-up\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: High-contrast monochrome daylight isolates the black hand against the chief's throat, with restrained gold confined to any sliver of the figure's fading eye left at the frame edge.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same daylight exterior look and the ongoing bloodied aftermath in front of the wooden shed. Exclude the full spear tableau and all unrelated bodies from the tight frame; retain only the gripping hand and the elder's neck.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The mortally wounded black figure still has the white-antler spear through its chest as it grips the village chief’s neck.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 조선인 촌주 (Korean man in his mid-50s, dark graying hair, weathered mature face); 그분(검은 연기 형체) (smoke-black humanoid head, two bent horns, vaguely human face shape) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "검은 형체의 팔이 화면 좌상단에서 뻗어 나와 우측 촌주의 목(턱 아래 인후)을 정확히 감싼다. 손가락이 목을 향해 있고 촌주의 시선은 위로 젖혀져 있어 조르는 방향은 맞다. 다만 프레임은 클로즈업이 아니라 미디엄/와이드에 가까워, 창고 전체·백골 뿔창·시신 여러 구가 배경에 크게 노출된다. 두 얼굴이 상단 프레임 밖으로 벗어나야 한다는 지시와 달리 촌주 얼굴 전체가 완전히 보인다.",
    "built_space": "배경은 참조의 목조 소금창고와 갈대밭, 흙바닥 위 자루형 시신들이 그대로 재현되어 위치 락은 충족. 그러나 '배제하라'고 명시된 흰 뿔창 타블로와 무관한 시신들이 프레임 안에 그대로 남아 있다.",
    "entities": "촌주는 50대 중반 회끗한 수염·짙은 두건·감청 도포로 참조와 일치. 검은 손은 길쭉하고 발톱이 있으나 표면이 연기라기보다 마른 인간 피부에 가깝고, 좌상단 눈의 금빛 슬리버는 존재. 가슴을 관통한 흰 뿔창(운반 상태)은 형체 몸이 프레임 밖이라 확인 불가.",
    "hard_violations": [],
    "physics": "촌주는 무릎 꿇거나 상체가 뒤로 젖혀진 자세로 하반신이 화면 밖 지면에 닿아 있어 지지가 설명된다. 검은 팔은 어깨·몸통으로 이어져 좌상단 형체에 연결, 공중 부유 없음. 목을 조이는 손가락의 압박 흔적은 약하지만 접촉은 성립."
   },
   {
    "label": "B",
    "direction": "검은 손/팔이 촌주의 목을 감싸며 조르는 형태로 뻗어 있고, 손목이 화면 중앙 대각선을 가로지른다. 촌주 얼굴은 위로 젖혀져 상단으로 잘려 부분만 보이고, 금빛 눈은 좌상단 프레임 가장자리에 있어 지시에 부합. 다만 실제 손 형태가 불분명하게 연기·천처럼 뭉개져 '손가락이 목을 쥔' 방향성이 약하다.",
    "built_space": "배경에 창고, 갈대, 자루 시신이 재현되나 프레임 우측 중경에 흰 사슴뿔 창끝이 촌주 어깨 뒤에서 떠 있는 형태로 나타나 배제 지시를 어기고, 실제 크기·거리 대비 과대 확대되어 전경 요소처럼 보인다. 또한 좌우 양쪽 큰 검은 여백(레터박스 유사)이 있어 화면이 패널처럼 나뉜 인상.",
    "entities": "촌주 얼굴 상부·수염은 참조와 유사. 검은 형체는 손이라기보다 목을 휘감은 연기 덩어리/천 형상으로, '길쭉한 검은 손'의 해부학이 성립하지 않는다. 금빛 균열과 눈은 참조와 유사.",
    "hard_violations": [
     "흰 뿔창이 촌주 어깨 뒤 허공에 아무 지지 없이 박혀 떠 있는 상태 — 관통한 몸도 잡은 손도 보이지 않아 지지체 없음",
     "화면 좌우에 검은 띠가 삽입되어 이미지가 패널/콜라주처럼 분할됨",
     "목을 감싼 검은 부분에 손가락·손 구조가 없어 해부학적으로 '쥐는 손'이 성립하지 않음"
    ],
    "physics": "촌주의 상체는 프레임 하단으로 이어져 서 있거나 무릎 꿇은 상태로 추정되어 지지는 있으나, 뿔창은 어떤 몸통이나 손에도 연결되지 않은 채 공중에 떠 있다."
   },
   {
    "label": "C",
    "direction": "길쭉하고 뼈마디가 도드라진 검은 손이 촌주의 목젖 아래를 정면으로 강하게 틀어쥐고 있으며, 손가락 다섯이 목을 감싼 접촉면이 뚜렷하다. 촌주의 머리는 뒤로 젖혀져 눈·이마 대부분이 상단 프레임 밖으로 잘리고, 형체의 팔목이 화면 중앙 대각선을 좌하단으로 가로지른다 — 지시된 구도와 가장 일치.",
    "built_space": "배경 목조 창고 판벽, 문짝, 갈대, 흙바닥 위 자루 시신 한 구가 참조 그대로이며 원근상 작게 물러나 있다. 흰 뿔은 중경에 작게 보여 배경 요소 크기 규칙을 지킨다(엄밀히는 배제 지시에 걸치는 잔존 요소).",
    "entities": "촌주: 50대 중반, 회백색 수염, 검은 두건, 감청/검정 도포 — 참조 인물과 일치. 검은 손: 연기 질감의 검은 표면에 금빛 균열이 흐르는 참조 형체의 재질과 일치, 길쭉한 손가락. 금빛은 균열에만 제한적으로 남아 고대비 흑백 톤 유지.",
    "hard_violations": [],
    "physics": "촌주의 몸통과 어깨가 화면 하단으로 이어져 무게가 실린 상태이며, 검은 팔은 좌하단으로 이어져 프레임 밖 몸통에 연결된다. 목의 피부가 손가락에 눌려 밀리는 변형이 보여 '강하게 틀어쥔' 순간의 힘이 물리적으로 읽힌다. 부유하는 물체 없음."
   }
  ],
  "totals": {
   "A": 5,
   "B": 2,
   "C": 8
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "목을 쥔 방향은 맞지만 클로즈업이 아닌 미디엄/와이드로 창고 전경·뿔창 타블로·시신들을 그대로 노출해 '배제' 지시와 '두 얼굴은 상단 프레임 밖' 구도 지시를 어겼고, 검은 손 질감도 연기가 아닌 마른 인간 피부에 가깝다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "흰 뿔창이 아무 몸통·손에도 연결되지 않은 채 허공에 떠 있고 좌우 검은 띠로 화면이 패널처럼 분할되며, 목을 감싼 검은 부분에 손가락 구조가 없어 '길쭉한 검은 손이 목을 틀어쥔' 핵심 동작 자체가 성립하지 않는다."
   },
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "목 높이 근접 클로즈업, 젖혀진 목과 뻗은 손목이 중앙 대각선을 가로지르고 두 얼굴 대부분이 상단 밖으로 잘린 지시 구도를 정확히 실현했으며, 피부가 눌리는 압박 변형과 연기·금빛 균열 재질이 참조 형체와 일치한다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S3sh6_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 조선인 촌주: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1414509>"
   },
   {
    "label": "CHARACTER REFERENCE — 그분(검은 연기 형체): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1138509>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "촌주 뒤편 배경의 자루 위로 솟아 있는 흰색 뿔은 설정상 화면 밖에 있는 검은 형체의 가슴에 박혀 있어야 하므로, 배경 바닥에 있어서는 안 됩니다.",
     "fix_en": "Completely remove the white antler sprouting from behind the burlap sack in the left background, filling the space with the continuation of the dry reeds and muddy ground. Preserve the village chief, his pose and clothing, the black hand and arm gripping his neck, the wooden shed, the burlap sacks on the ground, the lighting, and the exact camera framing.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Completely remove the white antler sprouting from behind the burlap sack in the left background, filling the space with the continuation of the dry reeds and muddy ground. Preserve the village chief, his pose and clothing, the black hand and arm gripping his neck, the wooden shed, the burlap sacks on the ground, the lighting, and the exact camera framing.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "동일 구도이나 촌주 얼굴이 상단에 더 많이 남아 '얼굴 대부분은 상단 프레임 밖' 지시에서 벗어나고, 배경에 흰 뿔창이 크게 드러나 '창 타블로 제외' 지시를 어긴다.  ★위반: [gpt] 타이트 프레임에서 제외하라고 명시된 관련 없는 쓰러진 병사들의 몸 여러 구가 추가 인물로 보인다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "손이 목을 쥔 대각 구성과 목 압박 물리가 동일하게 성립하면서, 얼굴 크롭이 상단 프레임 밖으로 더 정확히 잘리고 뿔창·불필요 요소를 프레임에서 배제해 '타이트 클로즈업, 창 타블로 제외' 지시를 더 충실히 만족한다.  ★위반: [gpt] 타이트 프레임에서 제외하라고 명시된 관련 없는 쓰러진 병사들의 몸 여러 구가 추가 인물로 보인다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.583,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.333,
      "B": 1.75
     },
     "violations": {
      "A": [
       "[gpt] 타이트 프레임에서 제외하라고 명시된 관련 없는 쓰러진 병사들의 몸 여러 구가 추가 인물로 보인다."
      ],
      "B": [
       "[gpt] 타이트 프레임에서 제외하라고 명시된 관련 없는 쓰러진 병사들의 몸 여러 구가 추가 인물로 보인다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "동일 구도이나 촌주 얼굴이 상단에 더 많이 남아 '얼굴 대부분은 상단 프레임 밖' 지시에서 벗어나고, 배경에 흰 뿔창이 크게 드러나 '창 타블로 제외' 지시를 어긴다.  ★위반: [gpt] 타이트 프레임에서 제외하라고 명시된 관련 없는 쓰러진 병사들의 몸 여러 구가 추가 인물로 보인다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "손이 목을 쥔 대각 구성과 목 압박 물리가 동일하게 성립하면서, 얼굴 크롭이 상단 프레임 밖으로 더 정확히 잘리고 뿔창·불필요 요소를 프레임에서 배제해 '타이트 클로즈업, 창 타블로 제외' 지시를 더 충실히 만족한다.  ★위반: [gpt] 타이트 프레임에서 제외하라고 명시된 관련 없는 쓰러진 병사들의 몸 여러 구가 추가 인물로 보인다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "프레이밍·손의 파지·인물 동일성은 충족하나, CARRIED STATE(백각 창)의 흔적이 전혀 없고 배경 대비가 상대적으로 채도가 높아 '고대비 모노크롬' 지시에서 약간 밀린다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 구도와 파지를 유지하면서 배경에 백각 창의 흰 뿔을 실제 스케일로 되살려 CARRIED STATE를 충족하고, 탈색된 고대비 톤과 절제된 금빛이 조명 지시에 더 맞는다 — 변경 범위가 배경 요소와 톤에 국한되어 손실이 작다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.75,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.75,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "프레이밍·손의 파지·인물 동일성은 충족하나, CARRIED STATE(백각 창)의 흔적이 전혀 없고 배경 대비가 상대적으로 채도가 높아 '고대비 모노크롬' 지시에서 약간 밀린다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 구도와 파지를 유지하면서 배경에 백각 창의 흰 뿔을 실제 스케일로 되살려 CARRIED STATE를 충족하고, 탈색된 고대비 톤과 절제된 금빛이 조명 지시에 더 맞는다 — 변경 범위가 배경 요소와 톤에 국한되어 손실이 작다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 15
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S3sh6"
  }
 },
 "S5sh1::bgfirst_bg": {
  "input_fingerprint": "8a4a7ba26c4feb48",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 캔 음료가 가득 찬 편의점 냉장고 뒤편 빈 공간에서 파란색 조끼를 입은 김지후(강수리영)가 유통기한이 지난 도시락을 입에 문 채 멍하게 씹고 있는 모습.\n\nLOCATION (lock): The narrow staff-side space behind a convenience-store beverage refrigerator, enclosed by rows of stocked cans and glass cooler doors.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Begin behind the refrigerator at chest height and dolly toward 김지후(강수리영) through narrow gaps in the densely stocked cans, keeping the converging rows as a layered foreground frame. Hold him in the center midground, leaning over the expired boxed meal with the blue vest and vacant chewing expression readable, while the disposal basket remains low in the surrounding confined space.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 가득 진열된 캔 음료 (냉장고 안에 꽉 들어차 있음) — The camera faces the sides and gaps of the tightly packed rows, creating a narrow visual corridor toward 김지후; used as Forms repeated foreground occlusion through which the dolly-in discovers 김지후; 폐기처분 음식 바구니 (폐기 음식이 가득 담겨 있음); used as Anchors the discarded meal supply in the lower portion of the confined work area; 냉장고 뒤편 빈 공간 (김지후가 늦은 저녁을 먹는 중) — The camera looks along the confined space toward 김지후 through the stocked refrigerator; used as Defines the rear empty space occupied by 김지후.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient convenience-store lighting retains the blue vest and tired complexion without introducing a visible light fixture.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 캔 음료가 가득 찬 편의점 냉장고 뒤편 빈 공간에서 파란색 조끼를 입은 김지후(강수리영)가 유통기한이 지난 도시락을 입에 문 채 멍하게 씹고 있는 모습.\n\nLOCATION (lock): The narrow staff-side space behind a convenience-store beverage refrigerator, enclosed by rows of stocked cans and glass cooler doors.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Begin behind the refrigerator at chest height and dolly toward 김지후(강수리영) through narrow gaps in the densely stocked cans, keeping the converging rows as a layered foreground frame. Hold him in the center midground, leaning over the expired boxed meal with the blue vest and vacant chewing expression readable, while the disposal basket remains low in the surrounding confined space.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 가득 진열된 캔 음료 (냉장고 안에 꽉 들어차 있음) — The camera faces the sides and gaps of the tightly packed rows, creating a narrow visual corridor toward 김지후; used as Forms repeated foreground occlusion through which the dolly-in discovers 김지후; 폐기처분 음식 바구니 (폐기 음식이 가득 담겨 있음); used as Anchors the discarded meal supply in the lower portion of the confined work area; 냉장고 뒤편 빈 공간 (김지후가 늦은 저녁을 먹는 중) — The camera looks along the confined space toward 김지후 through the stocked refrigerator; used as Defines the rear empty space occupied by 김지후.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient convenience-store lighting retains the blue vest and tired complexion without introducing a visible light fixture.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S5sh1__bgfirst_bg.png",
  "asset_id": "bc8b3072-0fdb-4148-b7fb-f8979fa89ae3",
  "input_asset_ids": [
   "e5344510-caf6-4451-b32a-98d43f16e35d",
   "498675e2-b0d6-4991-86f0-9567ecdd0214"
  ]
 },
 "S5sh1": {
  "input_fingerprint": "103ad5cabfb83465",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 캔 음료가 가득 찬 편의점 냉장고 뒤편 빈 공간에서 파란색 조끼를 입은 김지후(강수리영)가 유통기한이 지난 도시락을 입에 문 채 멍하게 씹고 있는 모습.\n\nLOCATION (lock): The narrow staff-side space behind a convenience-store beverage refrigerator, enclosed by rows of stocked cans and glass cooler doors. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Begin behind the refrigerator at chest height and dolly toward 김지후(강수리영) through narrow gaps in the densely stocked cans, keeping the converging rows as a layered foreground frame. Hold him in the center midground, leaning over the expired boxed meal with the blue vest and vacant chewing expression readable, while the disposal basket remains low in the surrounding confined space.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 가득 진열된 캔 음료 (냉장고 안에 꽉 들어차 있음) — The camera faces the sides and gaps of the tightly packed rows, creating a narrow visual corridor toward 김지후; used as Forms repeated foreground occlusion through which the dolly-in discovers 김지후; 폐기처분 음식 바구니 (폐기 음식이 가득 담겨 있음); used as Anchors the discarded meal supply in the lower portion of the confined work area; 냉장고 뒤편 빈 공간 (김지후가 늦은 저녁을 먹는 중) — The camera looks along the confined space toward 김지후 through the stocked refrigerator; used as Defines the rear empty space occupied by 김지후.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient convenience-store lighting retains the blue vest and tired complexion without introducing a visible light fixture.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu continues to wear her white animal-bone, spearhead-shaped necklace with the same blue work vest.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 김지후(강수리영) right now, so 김지후(강수리영)'s hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 김지후(강수리영): its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 캔 음료가 가득 찬 편의점 냉장고 뒤편 빈 공간에서 파란색 조끼를 입은 김지후(강수리영)가 유통기한이 지난 도시락을 입에 문 채 멍하게 씹고 있는 모습.\n\nLOCATION (lock): The narrow staff-side space behind a convenience-store beverage refrigerator, enclosed by rows of stocked cans and glass cooler doors. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Begin behind the refrigerator at chest height and dolly toward 김지후(강수리영) through narrow gaps in the densely stocked cans, keeping the converging rows as a layered foreground frame. Hold him in the center midground, leaning over the expired boxed meal with the blue vest and vacant chewing expression readable, while the disposal basket remains low in the surrounding confined space.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 가득 진열된 캔 음료 (냉장고 안에 꽉 들어차 있음) — The camera faces the sides and gaps of the tightly packed rows, creating a narrow visual corridor toward 김지후; used as Forms repeated foreground occlusion through which the dolly-in discovers 김지후; 폐기처분 음식 바구니 (폐기 음식이 가득 담겨 있음); used as Anchors the discarded meal supply in the lower portion of the confined work area; 냉장고 뒤편 빈 공간 (김지후가 늦은 저녁을 먹는 중) — The camera looks along the confined space toward 김지후 through the stocked refrigerator; used as Defines the rear empty space occupied by 김지후.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient convenience-store lighting retains the blue vest and tired complexion without introducing a visible light fixture.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu continues to wear her white animal-bone, spearhead-shaped necklace with the same blue work vest.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 김지후(강수리영) right now, so 김지후(강수리영)'s hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 김지후(강수리영): its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 캔 음료가 가득 찬 편의점 냉장고 뒤편 빈 공간에서 파란색 조끼를 입은 김지후(강수리영)가 유통기한이 지난 도시락을 입에 문 채 멍하게 씹고 있는 모습.\n\nLOCATION (lock): The narrow staff-side space behind a convenience-store beverage refrigerator, enclosed by rows of stocked cans and glass cooler doors. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Begin behind the refrigerator at chest height and dolly toward 김지후(강수리영) through narrow gaps in the densely stocked cans, keeping the converging rows as a layered foreground frame. Hold him in the center midground, leaning over the expired boxed meal with the blue vest and vacant chewing expression readable, while the disposal basket remains low in the surrounding confined space.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 가득 진열된 캔 음료 (냉장고 안에 꽉 들어차 있음) — The camera faces the sides and gaps of the tightly packed rows, creating a narrow visual corridor toward 김지후; used as Forms repeated foreground occlusion through which the dolly-in discovers 김지후; 폐기처분 음식 바구니 (폐기 음식이 가득 담겨 있음); used as Anchors the discarded meal supply in the lower portion of the confined work area; 냉장고 뒤편 빈 공간 (김지후가 늦은 저녁을 먹는 중) — The camera looks along the confined space toward 김지후 through the stocked refrigerator; used as Defines the rear empty space occupied by 김지후.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient convenience-store lighting retains the blue vest and tired complexion without introducing a visible light fixture.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu continues to wear her white animal-bone, spearhead-shaped necklace with the same blue work vest.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 김지후(강수리영) right now, so 김지후(강수리영)'s hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 김지후(강수리영): its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S5sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S5sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:904317>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L05B03.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:904317>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "캔 통로 전경과 인물 중앙 배치, 인물 유사도는 좋으나 '유통기한 지난 도시락'이 종이 포장 샌드위치로 바뀌었고, 도시락 위로 숙인 자세와 하단 폐기 바구니 앵커가 약하며 후면 스태프 공간이라기보다 매장 통로로 읽힌다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "도시락 트레이를 든 손·숙인 상체·멍한 씹는 표정, 하단 중앙의 레퍼런스와 동일한 '폐기용' 바구니, 녹슨 금속 후면 벽까지 로케이션 락과 샷 텍스트를 가장 충실히 구현했다(얼굴 유사도와 뼈 펜던트 형태는 소폭 이탈)."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.5,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.5,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "캔 통로 전경과 인물 중앙 배치, 인물 유사도는 좋으나 '유통기한 지난 도시락'이 종이 포장 샌드위치로 바뀌었고, 도시락 위로 숙인 자세와 하단 폐기 바구니 앵커가 약하며 후면 스태프 공간이라기보다 매장 통로로 읽힌다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "도시락 트레이를 든 손·숙인 상체·멍한 씹는 표정, 하단 중앙의 레퍼런스와 동일한 '폐기용' 바구니, 녹슨 금속 후면 벽까지 로케이션 락과 샷 텍스트를 가장 충실히 구현했다(얼굴 유사도와 뼈 펜던트 형태는 소폭 이탈)."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "샷 텍스트의 핵심(도시락 용기를 들고 입에 문 채 멍하게 씹기), 수렴하는 캔 열의 전경 프레이밍, 하단의 폐기용 바구니, 저조도 야간 무드를 모두 충족 — 뼈 목걸이 형태가 창끝형이 아니고 좌측 유리 반사가 다소 과한 점이 감점.  ★위반: [gpt] 폐기 바구니와 식품 포장에 선명한 한글 및 상품 문구가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "인물 동일성과 목걸이 형태는 더 근접하나, 중심 소품이 도시락이 아닌 샌드위치이고 폐기 바구니가 하단 앵커가 아닌 배경으로 밀렸으며 조명이 야간 저조도라기보다 밝아 지정 사양에서 벗어난다.  ★위반: [gpt] 캔과 의류의 상품 표기 형태가 화면에 남아 있어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.286
     },
     "adjusted": {
      "A": 1.75,
      "B": 1.036
     },
     "violations": {
      "A": [
       "[gpt] 폐기 바구니와 식품 포장에 선명한 한글 및 상품 문구가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
      ],
      "B": [
       "[gpt] 캔과 의류의 상품 표기 형태가 화면에 남아 있어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "샷 텍스트의 핵심(도시락 용기를 들고 입에 문 채 멍하게 씹기), 수렴하는 캔 열의 전경 프레이밍, 하단의 폐기용 바구니, 저조도 야간 무드를 모두 충족 — 뼈 목걸이 형태가 창끝형이 아니고 좌측 유리 반사가 다소 과한 점이 감점.  ★위반: [gpt] 폐기 바구니와 식품 포장에 선명한 한글 및 상품 문구가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "인물 동일성과 목걸이 형태는 더 근접하나, 중심 소품이 도시락이 아닌 샌드위치이고 폐기 바구니가 하단 앵커가 아닌 배경으로 밀렸으며 조명이 야간 저조도라기보다 밝아 지정 사양에서 벗어난다.  ★위반: [gpt] 캔과 의류의 상품 표기 형태가 화면에 남아 있어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 15
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 11,
   "B": 15
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "캔 통로 전경과 인물 중앙 배치, 인물 유사도는 좋으나 '유통기한 지난 도시락'이 종이 포장 샌드위치로 바뀌었고, 도시락 위로 숙인 자세와 하단 폐기 바구니 앵커가 약하며 후면 스태프 공간이라기보다 매장 통로로 읽힌다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "도시락 트레이를 든 손·숙인 상체·멍한 씹는 표정, 하단 중앙의 레퍼런스와 동일한 '폐기용' 바구니, 녹슨 금속 후면 벽까지 로케이션 락과 샷 텍스트를 가장 충실히 구현했다(얼굴 유사도와 뼈 펜던트 형태는 소폭 이탈)."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L05B03.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:904317>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "왼쪽 냉장고 유리문에 현재 지후의 자세 및 시선과 전혀 일치하지 않는 불가능한 얼굴과 팔의 반사상(또는 가상의 인물)이 비쳐 있습니다.",
     "fix_en": "Remove the impossible face and arm reflection from the glass door on the far left, replacing them with the normal dark interior and reflections of the beverage shelves. Preserve Jihu, her exact pose and expression, her clothing, the lunchbox she is holding, the food in her mouth, the foreground basket, and the rest of the convenience store environment exactly as they are.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "목걸이 펜던트가 지시된 '창촉(spearhead) 모양'이 아니라 만화적인 개뼈다귀 모양으로 그려졌습니다.",
     "fix_en": "Redraw the white pendant on the necklace to be a sharp, pointed spearhead shape. Preserve Jihu's face, her exact expression, the food in her mouth, her hands, the lunchbox, her blue vest, and the entire surrounding background exactly as they are.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the impossible face and arm reflection from the glass door on the far left, replacing them with the normal dark interior and reflections of the beverage shelves. Preserve Jihu, her exact pose and expression, her clothing, the lunchbox she is holding, the food in her mouth, the foreground basket, and the rest of the convenience store environment exactly as they are.\n- Redraw the white pendant on the necklace to be a sharp, pointed spearhead shape. Preserve Jihu's face, her exact expression, the food in her mouth, her hands, the lunchbox, her blue vest, and the entire surrounding background exactly as they are.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "구도·조명·장소·인물 모두 계약을 충족하지만, CARRIED STATE가 명시한 '흰 동물뼈 창끝 모양 펜던트'가 개뼈다귀 형태로 렌더되어 지속성 조항을 어긴다.  ★위반: [gpt] 포장지와 바구니에 ‘폐기용’을 비롯한 식별 가능한 문자와 상품 문구가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "A와 동일한 프레이밍·인물·소품·조명을 유지하면서 오직 목걸이만 계약대로 창끝 모양 뼈 펜던트로 수정했고, 얼굴·머리·조끼·손동작·폐기 바구니 등 계약이 요구하지 않은 요소는 사실상 변화가 없어 수리 비용이 거의 없다.  ★위반: [gpt] 포장지와 바구니에 ‘폐기용’을 비롯한 식별 가능한 문자와 상품 문구가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.732,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.482,
      "B": 1.75
     },
     "violations": {
      "A": [
       "[gpt] 포장지와 바구니에 ‘폐기용’을 비롯한 식별 가능한 문자와 상품 문구가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
      ],
      "B": [
       "[gpt] 포장지와 바구니에 ‘폐기용’을 비롯한 식별 가능한 문자와 상품 문구가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "구도·조명·장소·인물 모두 계약을 충족하지만, CARRIED STATE가 명시한 '흰 동물뼈 창끝 모양 펜던트'가 개뼈다귀 형태로 렌더되어 지속성 조항을 어긴다.  ★위반: [gpt] 포장지와 바구니에 ‘폐기용’을 비롯한 식별 가능한 문자와 상품 문구가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "A와 동일한 프레이밍·인물·소품·조명을 유지하면서 오직 목걸이만 계약대로 창끝 모양 뼈 펜던트로 수정했고, 얼굴·머리·조끼·손동작·폐기 바구니 등 계약이 요구하지 않은 요소는 사실상 변화가 없어 수리 비용이 거의 없다.  ★위반: [gpt] 포장지와 바구니에 ‘폐기용’을 비롯한 식별 가능한 문자와 상품 문구가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후의 시선은 카메라 아래쪽/약간 정면을 향해 초점 없이 멍하게 풀려 있고, 얼굴은 정면을 향한다. 도시락 트레이는 두 손 사이에 들려 몸 쪽(자기 눈 아래)을 향해 열려 있어 실제 식사 방향과 일치한다. 왼쪽 유리문에는 본인의 반사상이 보이며, 반사 속 인물도 같은 방향으로 손을 뻗고 있어 카메라 위치에서 발생 가능한 상이다.",
      "built_space": "좌우로 캔·페트병이 빽빽한 냉장고 진열 열이 전경 프레임을 이루고, 중앙에 냉장고 뒤편 좁은 통로(회색 금속 벽, 배관)가 열려 있다. 하단에는 '폐기용' 검정 바구니에 폐기 도시락·컵라면이 가득 담겨 로케이션 사진의 오렌지/블랙 크레이트와 일치. 냉장고 열 수, 통로 폭, 시선 회랑 모두 로케이션과 부합. 좌측 유리 반사 1개(본인상)로 중복 인물 아님.",
      "entities": "20대 초반 한국 여성, 긴 검은 생머리, 파란 형광 조끼+회색 티셔츠 — 캐릭터 레퍼런스와 일치. 목에는 흰색 뼈/창끝(스피어헤드) 형태의 펜던트가 은색 체인에 걸려 있어 CARRIED STATE의 '동물뼈 창끝 모양 목걸이'와 부합. 도시락은 알루미늄/플라스틱 트레이의 실제 편의점 도시락, 입에는 씹던 음식 조각이 물려 있다.",
      "hard_violations": [],
      "physics": "여성은 통로 안에 서서(또는 살짝 기대어) 양손으로 도시락 트레이를 아래에서 받쳐 들고 있으며, 손·손목·팔뚝이 프레임 안에 있고 그립이 실제 사용법과 일치. 트레이는 두 손이 명확히 지지. 폐기 바구니는 바닥에 놓여 있다. 공중에 뜬 물체 없음. 입 안의 음식 조각은 씹는 동작과 일관."
     },
     {
      "label": "B",
      "direction": "시선·얼굴 방향은 A와 동일하게 정면 아래로 초점 없는 멍한 표정. 도시락 트레이도 자기 쪽으로 향해 있어 방향은 맞다. 다만 좌측 유리문 반사상에서 인물의 머리 부분이 어둡게 뭉개져 얼굴이 거의 식별되지 않고, 반사 팔만 남아 반사 논리가 A보다 흐릿하다.",
      "built_space": "냉장고 열·통로·폐기 바구니 배치는 A와 사실상 동일하며 로케이션과 부합. 전체적으로 이미지가 더 흐릿(소프트)하고 대비가 낮아 캔 라벨·선반 디테일이 뭉개져 있다. 좌측 반사 영역이 어둡게 처리되어 공간 정보가 줄었다.",
      "entities": "동일 인물, 긴 검은 머리, 파란 조끼. 그러나 목걸이가 '창끝 모양 흰 뼈'가 아니라 만화적인 개뼈다귀(양끝이 부푼 도그본) 형태로 바뀌어 CARRIED STATE의 스피어헤드 명세와 어긋나고 재질감도 플라스틱 스티커처럼 납작하다. 조끼의 반사 스트라이프와 로고 디테일도 흐려졌다.",
      "hard_violations": [],
      "physics": "양손이 트레이를 받쳐 들고 있어 지지는 동일하게 성립. 입에 문 음식도 동일. 다만 전반적 해상감 저하로 손가락 경계와 트레이 가장자리가 뭉개져 물성이 약하다. 공중 부양 등 위반은 없다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "목걸이가 계약대로 흰 뼈 창끝 형태이고, 냉장고 열·폐기 바구니·좁은 통로의 로케이션 락과 미디엄 샷 구도를 충실히 지키며 손 그립·표정·질감 묘사가 선명하다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "구도·장소·자세는 A와 동일하나 계약이 요구한 '창끝 모양 뼈 목걸이'를 만화적 개뼈다귀로 바꿔 CARRIED STATE를 위반하고, 전체 선명도·질감·반사 디테일까지 함께 잃어 수리 대비 손실이 크다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "목걸이가 계약대로 흰 뼈 창끝 형태이고, 냉장고 열·폐기 바구니·좁은 통로의 로케이션 락과 미디엄 샷 구도를 충실히 지키며 손 그립·표정·질감 묘사가 선명하다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "구도·장소·자세는 A와 동일하나 계약이 요구한 '창끝 모양 뼈 목걸이'를 만화적 개뼈다귀로 바꿔 CARRIED STATE를 위반하고, 전체 선명도·질감·반사 디테일까지 함께 잃어 수리 대비 손실이 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "김지후의 시선은 카메라 아래쪽/약간 정면을 향해 초점 없이 멍하게 풀려 있고, 얼굴은 정면을 향한다. 도시락 트레이는 두 손 사이에 들려 몸 쪽(자기 눈 아래)을 향해 열려 있어 실제 식사 방향과 일치한다. 왼쪽 유리문에는 본인의 반사상이 보이며, 반사 속 인물도 같은 방향으로 손을 뻗고 있어 카메라 위치에서 발생 가능한 상이다.",
      "built_space": "좌우로 캔·페트병이 빽빽한 냉장고 진열 열이 전경 프레임을 이루고, 중앙에 냉장고 뒤편 좁은 통로(회색 금속 벽, 배관)가 열려 있다. 하단에는 '폐기용' 검정 바구니에 폐기 도시락·컵라면이 가득 담겨 로케이션 사진의 오렌지/블랙 크레이트와 일치. 냉장고 열 수, 통로 폭, 시선 회랑 모두 로케이션과 부합. 좌측 유리 반사 1개(본인상)로 중복 인물 아님.",
      "entities": "20대 초반 한국 여성, 긴 검은 생머리, 파란 형광 조끼+회색 티셔츠 — 캐릭터 레퍼런스와 일치. 목에는 흰색 뼈/창끝(스피어헤드) 형태의 펜던트가 은색 체인에 걸려 있어 CARRIED STATE의 '동물뼈 창끝 모양 목걸이'와 부합. 도시락은 알루미늄/플라스틱 트레이의 실제 편의점 도시락, 입에는 씹던 음식 조각이 물려 있다.",
      "hard_violations": [],
      "physics": "여성은 통로 안에 서서(또는 살짝 기대어) 양손으로 도시락 트레이를 아래에서 받쳐 들고 있으며, 손·손목·팔뚝이 프레임 안에 있고 그립이 실제 사용법과 일치. 트레이는 두 손이 명확히 지지. 폐기 바구니는 바닥에 놓여 있다. 공중에 뜬 물체 없음. 입 안의 음식 조각은 씹는 동작과 일관."
     },
     {
      "label": "A",
      "direction": "시선·얼굴 방향은 A와 동일하게 정면 아래로 초점 없는 멍한 표정. 도시락 트레이도 자기 쪽으로 향해 있어 방향은 맞다. 다만 좌측 유리문 반사상에서 인물의 머리 부분이 어둡게 뭉개져 얼굴이 거의 식별되지 않고, 반사 팔만 남아 반사 논리가 A보다 흐릿하다.",
      "built_space": "냉장고 열·통로·폐기 바구니 배치는 A와 사실상 동일하며 로케이션과 부합. 전체적으로 이미지가 더 흐릿(소프트)하고 대비가 낮아 캔 라벨·선반 디테일이 뭉개져 있다. 좌측 반사 영역이 어둡게 처리되어 공간 정보가 줄었다.",
      "entities": "동일 인물, 긴 검은 머리, 파란 조끼. 그러나 목걸이가 '창끝 모양 흰 뼈'가 아니라 만화적인 개뼈다귀(양끝이 부푼 도그본) 형태로 바뀌어 CARRIED STATE의 스피어헤드 명세와 어긋나고 재질감도 플라스틱 스티커처럼 납작하다. 조끼의 반사 스트라이프와 로고 디테일도 흐려졌다.",
      "hard_violations": [],
      "physics": "양손이 트레이를 받쳐 들고 있어 지지는 동일하게 성립. 입에 문 음식도 동일. 다만 전반적 해상감 저하로 손가락 경계와 트레이 가장자리가 뭉개져 물성이 약하다. 공중 부양 등 위반은 없다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S5sh1__bgfirst_bg.png",
   "bg_asset_id": "bc8b3072-0fdb-4148-b7fb-f8979fa89ae3",
   "bg_record_key": "S5sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S7sh5::bgfirst_bg": {
  "input_fingerprint": "944fd87561969513",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 다급한 표정의 강민숙(홍은미)이 돌아서려는 정인우의 팔뚝을 꽉 움켜쥔 채 힘을 주고 있는 찰나.\n\nLOCATION (lock): An open parking area outside the police station in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track close at upper-torso height along the side of the pair as 정인우 turns away, holding a slightly wider close composition that keeps 강민숙(홍은미)’s urgent face above her gripping hand. Place her in the left midground and his arrested turning torso on the right, with the clenched grip forming the central diagonal that visibly stops his escape.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 강민숙(홍은미) in the middle-left of the frame, midground, reaches for 정인우’s turning arm; 정인우 in the middle-right of the frame, midground, moves toward away from 강민숙 before being stopped.\n- KEY BACKGROUND ELEMENTS: 경찰서 주차장 (낮, 강민숙과 정인우가 대화하던 장소); used as Supplies restrained spatial context behind the close physical interruption.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight appropriate to the parking area is rendered with restrained contrast, preserving the recalled scene’s anxious realism.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 다급한 표정의 강민숙(홍은미)이 돌아서려는 정인우의 팔뚝을 꽉 움켜쥔 채 힘을 주고 있는 찰나.\n\nLOCATION (lock): An open parking area outside the police station in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track close at upper-torso height along the side of the pair as 정인우 turns away, holding a slightly wider close composition that keeps 강민숙(홍은미)’s urgent face above her gripping hand. Place her in the left midground and his arrested turning torso on the right, with the clenched grip forming the central diagonal that visibly stops his escape.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 강민숙(홍은미) in the middle-left of the frame, midground, reaches for 정인우’s turning arm; 정인우 in the middle-right of the frame, midground, moves toward away from 강민숙 before being stopped.\n- KEY BACKGROUND ELEMENTS: 경찰서 주차장 (낮, 강민숙과 정인우가 대화하던 장소); used as Supplies restrained spatial context behind the close physical interruption.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight appropriate to the parking area is rendered with restrained contrast, preserving the recalled scene’s anxious realism.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh5__bgfirst_bg.png",
  "asset_id": "9b5b40a9-478e-4d43-99a4-abf9dac90c8b",
  "input_asset_ids": [
   "595a0a6f-f89e-4dbd-8142-54edc21200c4",
   "2408e8fd-5d26-452a-8395-e3b913920e97"
  ]
 },
 "S7sh5": {
  "input_fingerprint": "ff4632c87eff9124",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 다급한 표정의 강민숙(홍은미)이 돌아서려는 정인우의 팔뚝을 꽉 움켜쥔 채 힘을 주고 있는 찰나.\n\nLOCATION (lock): An open parking area outside the police station in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track close at upper-torso height along the side of the pair as 정인우 turns away, holding a slightly wider close composition that keeps 강민숙(홍은미)’s urgent face above her gripping hand. Place her in the left midground and his arrested turning torso on the right, with the clenched grip forming the central diagonal that visibly stops his escape.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 강민숙(홍은미) in the middle-left of the frame, midground, reaches for 정인우’s turning arm; 정인우 in the middle-right of the frame, midground, moves toward away from 강민숙 before being stopped.\n- KEY BACKGROUND ELEMENTS: 경찰서 주차장 (낮, 강민숙과 정인우가 대화하던 장소); used as Supplies restrained spatial context behind the close physical interruption.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight appropriate to the parking area is rendered with restrained contrast, preserving the recalled scene’s anxious realism.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face); 강민숙(홍은미) (Korean woman in her late 40s, dark hair, refined narrow face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 다급한 표정의 강민숙(홍은미)이 돌아서려는 정인우의 팔뚝을 꽉 움켜쥔 채 힘을 주고 있는 찰나.\n\nLOCATION (lock): An open parking area outside the police station in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track close at upper-torso height along the side of the pair as 정인우 turns away, holding a slightly wider close composition that keeps 강민숙(홍은미)’s urgent face above her gripping hand. Place her in the left midground and his arrested turning torso on the right, with the clenched grip forming the central diagonal that visibly stops his escape.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 강민숙(홍은미) in the middle-left of the frame, midground, reaches for 정인우’s turning arm; 정인우 in the middle-right of the frame, midground, moves toward away from 강민숙 before being stopped.\n- KEY BACKGROUND ELEMENTS: 경찰서 주차장 (낮, 강민숙과 정인우가 대화하던 장소); used as Supplies restrained spatial context behind the close physical interruption.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight appropriate to the parking area is rendered with restrained contrast, preserving the recalled scene’s anxious realism.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face); 강민숙(홍은미) (Korean woman in her late 40s, dark hair, refined narrow face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 다급한 표정의 강민숙(홍은미)이 돌아서려는 정인우의 팔뚝을 꽉 움켜쥔 채 힘을 주고 있는 찰나.\n\nLOCATION (lock): An open parking area outside the police station in daylight. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track close at upper-torso height along the side of the pair as 정인우 turns away, holding a slightly wider close composition that keeps 강민숙(홍은미)’s urgent face above her gripping hand. Place her in the left midground and his arrested turning torso on the right, with the clenched grip forming the central diagonal that visibly stops his escape.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 강민숙(홍은미) in the middle-left of the frame, midground, reaches for 정인우’s turning arm; 정인우 in the middle-right of the frame, midground, moves toward away from 강민숙 before being stopped.\n- KEY BACKGROUND ELEMENTS: 경찰서 주차장 (낮, 강민숙과 정인우가 대화하던 장소); used as Supplies restrained spatial context behind the close physical interruption.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight appropriate to the parking area is rendered with restrained contrast, preserving the recalled scene’s anxious realism.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face); 강민숙(홍은미) (Korean woman in her late 40s, dark hair, refined narrow face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh5__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S7sh5.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:871658>"
    },
    {
     "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:862549>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L07B02.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:871658>"
    },
    {
     "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:862549>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "여자는 화면 왼쪽 중경, 남자는 오른쪽 중경으로 배치가 지시와 일치. 여자의 시선은 오른쪽 남자의 팔/등 쪽을 향하고, 남자는 오른쪽(카메라 오른편, 화면 밖)으로 얼굴을 돌려 돌아서려는 방향을 본다. 그러나 그녀의 두 손이 잡고 있는 '팔뚝'은 남자의 몸에서 나온 팔로 보이지 않는다 — 남자의 왼팔은 몸 옆에 따로 내려와 있고, 손에 쥔 소매는 남자 몸통 왼쪽에서 별도로 뻗어 나온 여분의 팔처럼 보인다. 즉 잡은 대상이 실제 남자의 팔뚝에 해부학적으로 연결되지 않는다.",
      "built_space": "경찰서 정면 유리 커튼월, 파란 띠, 금색 경찰 휘장, 볼라드, 화강암 인도 단, 주차선과 흰 SUV·검은 세단·순찰차 배치가 로케이션 사진과 매우 근접. 배경 요소 크기·거리 비율도 자연스럽다.",
      "entities": "여성: 40대 후반 한국인, 짙은 단발, 좁은 얼굴, 회색 더블 코트+검정 터틀넥 — 레퍼런스와 일치. 남성: 20대 한국인, 검은 머리, 감청 재킷+초록 스웨터 — 일치하나 레퍼런스의 회색 비니는 없음(치명적이진 않음). 제3의 인물 없음, 텍스트 없음.",
      "hard_violations": [
       "잡힌 팔이 남자의 몸통과 해부학적으로 연결되지 않음 — 몸 옆에 내려온 왼팔 외에 별도의 팔/소매가 존재하는 여분 팔다리(해부학적 불가능)"
      ],
      "physics": "두 사람 모두 지면에 발을 딛고 있고, 여자는 앞으로 체중을 실어 당기는 자세. 다만 붙잡힌 팔뚝은 지지·연결이 불명확해 공중에 떠 있는 소매처럼 보이며, 남자의 몸이 그 힘에 저항하는 텐션이 팔의 연결 부재로 무너진다."
     },
     {
      "label": "B",
      "direction": "여자는 화면 중앙 왼쪽 중경에서 오른손으로 남자의 왼팔 팔뚝(팔꿈치 아래)을 움켜쥐고, 시선은 남자 쪽/카메라 사선 쪽으로 다급하게 향한다. 남자는 오른쪽 중경에서 몸을 오른쪽으로 돌려 가려다 시선을 왼쪽 아래 잡힌 팔로 내리며 멈춰 선다. 잡힌 그립이 화면 중앙 대각선을 만들어 이탈을 막는 관계가 성립.",
      "built_space": "경찰서 외벽 흰 석재·파란 띠·금색 경찰 휘장·유리 자동문·볼라드·연석·주차선이 로케이션 사진과 일치. 좌측 순찰차와 검은 SUV는 참조 사진 배치와 유사. 배경 요소 스케일 적정, 전경화된 과대 오브젝트 없음.",
      "entities": "여성: 40대 후반 한국인, 어두운 단발, 좁고 단정한 얼굴, 회색 코트 — 레퍼런스 일치. 남성: 20대 중반 한국인, 회색 비니, 감청 블레이저, 초록 스웨터, 회색 팬츠 — 레퍼런스 의상까지 일치. 추가 인물·문자 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 아스팔트에 발을 딛고 무게가 실려 있다. 여자의 손은 남자의 팔뚝을 실제로 감싸 쥐고 있고, 남자의 팔은 어깨-팔꿈치-손목으로 자연스럽게 연결되며 손은 아래로 늘어져 제지당한 상태. 코트와 재킷의 주름·드레이프가 당기는 힘 방향과 일치."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "로케이션 재현과 좌우 배치는 좋으나, 강민숙이 쥔 소매가 정인우의 몸통에 연결되지 않는 여분의 팔처럼 보여 해부학적으로 불가능 — 샷의 핵심인 '팔뚝을 움켜쥔 순간'이 물리적으로 성립하지 않는다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "상반신 높이 측면 트래킹, 좌측 강민숙·우측 돌아서던 정인우, 중앙 대각선의 그립이 모두 지시대로 성립하고 인물 정체성·의상(비니 포함)과 경찰서 주차장 로케이션도 일치하며 해부학·물리 오류가 없다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "로케이션 재현과 좌우 배치는 좋으나, 강민숙이 쥔 소매가 정인우의 몸통에 연결되지 않는 여분의 팔처럼 보여 해부학적으로 불가능 — 샷의 핵심인 '팔뚝을 움켜쥔 순간'이 물리적으로 성립하지 않는다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "상반신 높이 측면 트래킹, 좌측 강민숙·우측 돌아서던 정인우, 중앙 대각선의 그립이 모두 지시대로 성립하고 인물 정체성·의상(비니 포함)과 경찰서 주차장 로케이션도 일치하며 해부학·물리 오류가 없다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "여자는 화면 왼쪽 중경, 남자는 오른쪽 중경으로 배치가 지시와 일치. 여자의 시선은 오른쪽 남자의 팔/등 쪽을 향하고, 남자는 오른쪽(카메라 오른편, 화면 밖)으로 얼굴을 돌려 돌아서려는 방향을 본다. 그러나 그녀의 두 손이 잡고 있는 '팔뚝'은 남자의 몸에서 나온 팔로 보이지 않는다 — 남자의 왼팔은 몸 옆에 따로 내려와 있고, 손에 쥔 소매는 남자 몸통 왼쪽에서 별도로 뻗어 나온 여분의 팔처럼 보인다. 즉 잡은 대상이 실제 남자의 팔뚝에 해부학적으로 연결되지 않는다.",
      "built_space": "경찰서 정면 유리 커튼월, 파란 띠, 금색 경찰 휘장, 볼라드, 화강암 인도 단, 주차선과 흰 SUV·검은 세단·순찰차 배치가 로케이션 사진과 매우 근접. 배경 요소 크기·거리 비율도 자연스럽다.",
      "entities": "여성: 40대 후반 한국인, 짙은 단발, 좁은 얼굴, 회색 더블 코트+검정 터틀넥 — 레퍼런스와 일치. 남성: 20대 한국인, 검은 머리, 감청 재킷+초록 스웨터 — 일치하나 레퍼런스의 회색 비니는 없음(치명적이진 않음). 제3의 인물 없음, 텍스트 없음.",
      "hard_violations": [
       "잡힌 팔이 남자의 몸통과 해부학적으로 연결되지 않음 — 몸 옆에 내려온 왼팔 외에 별도의 팔/소매가 존재하는 여분 팔다리(해부학적 불가능)"
      ],
      "physics": "두 사람 모두 지면에 발을 딛고 있고, 여자는 앞으로 체중을 실어 당기는 자세. 다만 붙잡힌 팔뚝은 지지·연결이 불명확해 공중에 떠 있는 소매처럼 보이며, 남자의 몸이 그 힘에 저항하는 텐션이 팔의 연결 부재로 무너진다."
     },
     {
      "label": "B",
      "direction": "여자는 화면 중앙 왼쪽 중경에서 오른손으로 남자의 왼팔 팔뚝(팔꿈치 아래)을 움켜쥐고, 시선은 남자 쪽/카메라 사선 쪽으로 다급하게 향한다. 남자는 오른쪽 중경에서 몸을 오른쪽으로 돌려 가려다 시선을 왼쪽 아래 잡힌 팔로 내리며 멈춰 선다. 잡힌 그립이 화면 중앙 대각선을 만들어 이탈을 막는 관계가 성립.",
      "built_space": "경찰서 외벽 흰 석재·파란 띠·금색 경찰 휘장·유리 자동문·볼라드·연석·주차선이 로케이션 사진과 일치. 좌측 순찰차와 검은 SUV는 참조 사진 배치와 유사. 배경 요소 스케일 적정, 전경화된 과대 오브젝트 없음.",
      "entities": "여성: 40대 후반 한국인, 어두운 단발, 좁고 단정한 얼굴, 회색 코트 — 레퍼런스 일치. 남성: 20대 중반 한국인, 회색 비니, 감청 블레이저, 초록 스웨터, 회색 팬츠 — 레퍼런스 의상까지 일치. 추가 인물·문자 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 아스팔트에 발을 딛고 무게가 실려 있다. 여자의 손은 남자의 팔뚝을 실제로 감싸 쥐고 있고, 남자의 팔은 어깨-팔꿈치-손목으로 자연스럽게 연결되며 손은 아래로 늘어져 제지당한 상태. 코트와 재킷의 주름·드레이프가 당기는 힘 방향과 일치."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "강민숙은 화면 좌중간에서 오른쪽의 정인우 쪽으로 시선과 몸을 향하고 있고, 정인우는 몸을 오른쪽(화면 밖)으로 돌리며 고개는 좌하방으로 돌려 그녀 쪽을 흘낏 본다. 그녀의 손은 그의 상완/팔뚝 부위를 잡고 있어 조준 대상은 맞다. 다만 시선 교환이 다소 어긋나 정인우가 그녀의 얼굴이 아니라 아래쪽 허공을 본다.",
      "built_space": "배경은 경찰서 주차장: 흰 주차선, 검은 SUV, 순찰차, 블루 밴드 외벽, 자동문, 경찰 휘장, 볼라드 — 로케이션 사진과 대체로 일치. 다만 좌측에 사다리차/트럭 같은 참조에 없는 차량이 추가되고 휘장 위치와 건물 비례가 다소 변형됨. 인물이 주차구획 위에 서 있어 공간 배치는 성립.",
      "entities": "정인우: 20대 중반 한국 남성, 회색 비니, 네이비 재킷, 초록 스웨터 — 캐릭터 참조와 일치(비니 포함). 강민숙: 40대 후반 한국 여성, 단발 흑발, 회색 더블 코트, 검은 터틀넥 — 참조와 일치. 제3의 인물 없음.",
      "hard_violations": [
       "정인우의 팔 개수 오류: 그녀가 잡고 있는 팔뚝 아래로 소매 밖에 또 하나의 주먹/손이 보이고, 화면 우측 아래에도 그의 오른손이 따로 있어 팔·손이 중복 렌더됨(해부학적 불가)"
      ],
      "physics": "두 인물 모두 지면에 발을 딛고 있고, 여성의 손은 남성 팔뚝을 실제로 감싼다. 그러나 그녀가 잡은 팔이 어느 어깨에서 나온 것인지 연결이 불명확하며, 그 아래 떠 있는 주먹은 어떤 어깨/소매와도 이어지지 않아 지지 근거가 없다."
     },
     {
      "label": "B",
      "direction": "강민숙은 중좌측에서 오른쪽의 정인우를 향해 몸을 기울이고 시선도 그의 등/팔에 고정. 정인우는 우측에서 몸을 오른쪽 화면 밖으로 돌리며 프로필로 앞을 응시 — 떠나려는 방향과 시선이 일치. 그녀의 두 손이 그의 왼팔뚝을 정확히 붙잡아 잡아채는 대각선이 화면 중앙을 가로지른다.",
      "built_space": "경찰서 정면 유리 자동문, 블루 밴드, 금색 경찰 휘장, 스테인리스 볼라드 열, 화강암 보도와 연석, 좌측 흰색 SUV·순찰차·검은 세단, 아스팔트 주차선 — 로케이션 참조의 요소와 배치가 매우 근접. 인물 스케일 대비 배경 요소 크기도 적정.",
      "entities": "정인우: 20대 중반 한국 남성, 검은 머리(비니 없음), 네이비 재킷+초록 스웨터+체크 셔츠 — 얼굴·건장은 참조와 유사하나 비니 미착용. 강민숙: 40대 후반 한국 여성, 검은 단발, 회색 코트, 검은 터틀넥 — 참조 일치. 추가 인물 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 지면에 확실히 발을 딛고 체중 이동이 보인다. 여성은 앞발에 체중을 실어 당기고, 남성은 반대 방향으로 몸을 틀며 팔이 뒤로 늘어나 제지되는 인장 관계가 성립. 잡힌 팔은 그의 어깨에서 자연스럽게 이어지고 손끝(주먹)까지 하나의 팔로 연결됨."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "비니 등 정인우 의상 재현은 좋으나, 붙잡힌 팔 아래로 어깨에 연결되지 않는 여분의 주먹/손이 중복 렌더돼 해부학적 불가 결함이 발생했고 시선 교환도 어긋나 제지의 긴장이 약하다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "상반신 높이 측면 트래킹, 좌중간 강민숙·우중간 돌아서는 정인우, 두 손으로 팔뚝을 움켜쥔 중앙 대각선이 지시대로 구현됐고 로케이션 요소도 일치 — 비니 미착용이 유일한 감점."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "비니 등 정인우 의상 재현은 좋으나, 붙잡힌 팔 아래로 어깨에 연결되지 않는 여분의 주먹/손이 중복 렌더돼 해부학적 불가 결함이 발생했고 시선 교환도 어긋나 제지의 긴장이 약하다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "상반신 높이 측면 트래킹, 좌중간 강민숙·우중간 돌아서는 정인우, 두 손으로 팔뚝을 움켜쥔 중앙 대각선이 지시대로 구현됐고 로케이션 요소도 일치 — 비니 미착용이 유일한 감점."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "강민숙은 화면 좌중간에서 오른쪽의 정인우 쪽으로 시선과 몸을 향하고 있고, 정인우는 몸을 오른쪽(화면 밖)으로 돌리며 고개는 좌하방으로 돌려 그녀 쪽을 흘낏 본다. 그녀의 손은 그의 상완/팔뚝 부위를 잡고 있어 조준 대상은 맞다. 다만 시선 교환이 다소 어긋나 정인우가 그녀의 얼굴이 아니라 아래쪽 허공을 본다.",
      "built_space": "배경은 경찰서 주차장: 흰 주차선, 검은 SUV, 순찰차, 블루 밴드 외벽, 자동문, 경찰 휘장, 볼라드 — 로케이션 사진과 대체로 일치. 다만 좌측에 사다리차/트럭 같은 참조에 없는 차량이 추가되고 휘장 위치와 건물 비례가 다소 변형됨. 인물이 주차구획 위에 서 있어 공간 배치는 성립.",
      "entities": "정인우: 20대 중반 한국 남성, 회색 비니, 네이비 재킷, 초록 스웨터 — 캐릭터 참조와 일치(비니 포함). 강민숙: 40대 후반 한국 여성, 단발 흑발, 회색 더블 코트, 검은 터틀넥 — 참조와 일치. 제3의 인물 없음.",
      "hard_violations": [
       "정인우의 팔 개수 오류: 그녀가 잡고 있는 팔뚝 아래로 소매 밖에 또 하나의 주먹/손이 보이고, 화면 우측 아래에도 그의 오른손이 따로 있어 팔·손이 중복 렌더됨(해부학적 불가)"
      ],
      "physics": "두 인물 모두 지면에 발을 딛고 있고, 여성의 손은 남성 팔뚝을 실제로 감싼다. 그러나 그녀가 잡은 팔이 어느 어깨에서 나온 것인지 연결이 불명확하며, 그 아래 떠 있는 주먹은 어떤 어깨/소매와도 이어지지 않아 지지 근거가 없다."
     },
     {
      "label": "A",
      "direction": "강민숙은 중좌측에서 오른쪽의 정인우를 향해 몸을 기울이고 시선도 그의 등/팔에 고정. 정인우는 우측에서 몸을 오른쪽 화면 밖으로 돌리며 프로필로 앞을 응시 — 떠나려는 방향과 시선이 일치. 그녀의 두 손이 그의 왼팔뚝을 정확히 붙잡아 잡아채는 대각선이 화면 중앙을 가로지른다.",
      "built_space": "경찰서 정면 유리 자동문, 블루 밴드, 금색 경찰 휘장, 스테인리스 볼라드 열, 화강암 보도와 연석, 좌측 흰색 SUV·순찰차·검은 세단, 아스팔트 주차선 — 로케이션 참조의 요소와 배치가 매우 근접. 인물 스케일 대비 배경 요소 크기도 적정.",
      "entities": "정인우: 20대 중반 한국 남성, 검은 머리(비니 없음), 네이비 재킷+초록 스웨터+체크 셔츠 — 얼굴·건장은 참조와 유사하나 비니 미착용. 강민숙: 40대 후반 한국 여성, 검은 단발, 회색 코트, 검은 터틀넥 — 참조 일치. 추가 인물 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 지면에 확실히 발을 딛고 체중 이동이 보인다. 여성은 앞발에 체중을 실어 당기고, 남성은 반대 방향으로 몸을 틀며 팔이 뒤로 늘어나 제지되는 인장 관계가 성립. 잡힌 팔은 그의 어깨에서 자연스럽게 이어지고 손끝(주먹)까지 하나의 팔로 연결됨."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 12
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "readings": [
   {
    "label": "A",
    "direction": "여자는 화면 왼쪽 중경, 남자는 오른쪽 중경으로 배치가 지시와 일치. 여자의 시선은 오른쪽 남자의 팔/등 쪽을 향하고, 남자는 오른쪽(카메라 오른편, 화면 밖)으로 얼굴을 돌려 돌아서려는 방향을 본다. 그러나 그녀의 두 손이 잡고 있는 '팔뚝'은 남자의 몸에서 나온 팔로 보이지 않는다 — 남자의 왼팔은 몸 옆에 따로 내려와 있고, 손에 쥔 소매는 남자 몸통 왼쪽에서 별도로 뻗어 나온 여분의 팔처럼 보인다. 즉 잡은 대상이 실제 남자의 팔뚝에 해부학적으로 연결되지 않는다.",
    "built_space": "경찰서 정면 유리 커튼월, 파란 띠, 금색 경찰 휘장, 볼라드, 화강암 인도 단, 주차선과 흰 SUV·검은 세단·순찰차 배치가 로케이션 사진과 매우 근접. 배경 요소 크기·거리 비율도 자연스럽다.",
    "entities": "여성: 40대 후반 한국인, 짙은 단발, 좁은 얼굴, 회색 더블 코트+검정 터틀넥 — 레퍼런스와 일치. 남성: 20대 한국인, 검은 머리, 감청 재킷+초록 스웨터 — 일치하나 레퍼런스의 회색 비니는 없음(치명적이진 않음). 제3의 인물 없음, 텍스트 없음.",
    "hard_violations": [
     "잡힌 팔이 남자의 몸통과 해부학적으로 연결되지 않음 — 몸 옆에 내려온 왼팔 외에 별도의 팔/소매가 존재하는 여분 팔다리(해부학적 불가능)"
    ],
    "physics": "두 사람 모두 지면에 발을 딛고 있고, 여자는 앞으로 체중을 실어 당기는 자세. 다만 붙잡힌 팔뚝은 지지·연결이 불명확해 공중에 떠 있는 소매처럼 보이며, 남자의 몸이 그 힘에 저항하는 텐션이 팔의 연결 부재로 무너진다."
   },
   {
    "label": "B",
    "direction": "여자는 화면 중앙 왼쪽 중경에서 오른손으로 남자의 왼팔 팔뚝(팔꿈치 아래)을 움켜쥐고, 시선은 남자 쪽/카메라 사선 쪽으로 다급하게 향한다. 남자는 오른쪽 중경에서 몸을 오른쪽으로 돌려 가려다 시선을 왼쪽 아래 잡힌 팔로 내리며 멈춰 선다. 잡힌 그립이 화면 중앙 대각선을 만들어 이탈을 막는 관계가 성립.",
    "built_space": "경찰서 외벽 흰 석재·파란 띠·금색 경찰 휘장·유리 자동문·볼라드·연석·주차선이 로케이션 사진과 일치. 좌측 순찰차와 검은 SUV는 참조 사진 배치와 유사. 배경 요소 스케일 적정, 전경화된 과대 오브젝트 없음.",
    "entities": "여성: 40대 후반 한국인, 어두운 단발, 좁고 단정한 얼굴, 회색 코트 — 레퍼런스 일치. 남성: 20대 중반 한국인, 회색 비니, 감청 블레이저, 초록 스웨터, 회색 팬츠 — 레퍼런스 의상까지 일치. 추가 인물·문자 없음.",
    "hard_violations": [],
    "physics": "두 인물 모두 아스팔트에 발을 딛고 무게가 실려 있다. 여자의 손은 남자의 팔뚝을 실제로 감싸 쥐고 있고, 남자의 팔은 어깨-팔꿈치-손목으로 자연스럽게 연결되며 손은 아래로 늘어져 제지당한 상태. 코트와 재킷의 주름·드레이프가 당기는 힘 방향과 일치."
   }
  ],
  "totals": {
   "A": 12,
   "B": 12
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "로케이션 재현과 좌우 배치는 좋으나, 강민숙이 쥔 소매가 정인우의 몸통에 연결되지 않는 여분의 팔처럼 보여 해부학적으로 불가능 — 샷의 핵심인 '팔뚝을 움켜쥔 순간'이 물리적으로 성립하지 않는다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "상반신 높이 측면 트래킹, 좌측 강민숙·우측 돌아서던 정인우, 중앙 대각선의 그립이 모두 지시대로 성립하고 인물 정체성·의상(비니 포함)과 경찰서 주차장 로케이션도 일치하며 해부학·물리 오류가 없다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L07B02.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:871658>"
   },
   {
    "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:862549>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "강민숙의 의상에서 캐릭터 레퍼런스에 명시된 회색 머플러가 누락되었습니다. (Major)",
     "fix_en": ""
    },
    {
     "issue_ko": "정인우의 스웨터 안쪽에 입은 셔츠가 레퍼런스의 단색 셔츠와 달리 체크무늬로 렌더링되었습니다. (Minor)",
     "fix_en": ""
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 5,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- \n- \n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "강민숙은 화면 오른쪽(정인우 쪽)을 향해 시선과 상체를 기울이고 두 손이 그의 오른팔 아래팔을 정면으로 움켜쥐고 있다. 정인우는 오른쪽(프레임 밖 출구 방향)으로 몸을 틀어 나가려는 자세이며 시선도 오른쪽 전방을 향한다. 스케치의 화살표 방향(여자 시선 오른쪽, 남자 이동 오른쪽)과 일치. 그립이 화면 중앙 대각선을 형성해 이탈을 저지하는 것이 읽힌다.",
      "built_space": "배경은 레퍼런스 경찰서 정면 그대로: 청색 캐노피와 금색 경찰 휘장, 유리 자동문, 스테인리스 볼라드 열, 화강암 보도 턱, 좌측 주차 구역의 검정 세단·흰색 SUV·순찰차. 유리문 안 반사와 세로 명패도 원본과 유사하게 유지. 볼라드 개수는 인물에 가려 일부만 보이지만 규모·원근이 자연스럽다.",
      "entities": "강민숙: 40대 후반 한국 여성, 검은 단발, 좁고 단정한 얼굴, 회색 울 코트+검정 터틀넥 — 레퍼런스와 일치(스카프·가방은 프레이밍상 없음, 감점 사유 아님). 정인우: 20대 중반 한국 남성, 검은 머리(비니 없음), 네이비 재킷+초록 스웨터+체크 셔츠 칼라 — 레퍼런스 의상과 일치. 제3의 인물 없음, 텍스트/워터마크 없음.",
      "hard_violations": [],
      "physics": "두 사람 모두 지면에 발을 딛고 있으며(하단 크롭이지만 체중 이동이 전방으로 기울어 있음), 여자는 앞발에 체중을 실어 뒤로 당기는 힘의 자세, 남자는 팔이 뒤로 끌려 어깨가 당겨진 상태. 여자의 두 손이 실제로 남자의 팔뚝을 감싸 접촉면과 옷 주름이 보인다. 공중에 뜬 물체 없음."
     },
     {
      "label": "B",
      "direction": "A와 거의 동일: 여자 시선·상체는 오른쪽 남자 쪽, 손은 그의 아래팔을 잡고, 남자는 오른쪽으로 돌아서며 시선은 오른쪽 전방. 방향성 자체는 문제 없음.",
      "built_space": "배경 유리 자동문 영역이 원본과 달라졌다. 원본의 투명한 유리(내부 반사·안내문·작은 경찰 마크)가 불투명한 흰 패널로 바뀌고, 그 위에 크게 확대된 두 개의 경찰 엠블럼과 판독 불가한 영문/한글 문구('SEOUL METROPOLITAN POLICE AGENCY' 유사 텍스트) 밴드가 문 전체에 붙었다. 즉 배경 픽스처가 변경되고 문자 요소가 추가됨. 볼라드·차량·캐노피는 동일.",
      "entities": "인물 동일: 강민숙(회색 코트, 검은 단발), 정인우(네이비 재킷, 초록 스웨터). 인물 측면에서는 A와 사실상 차이가 없다. 다만 배경에 계약이 금지한 텍스트/캡션성 표기가 다수 삽입되어 있다.",
      "hard_violations": [
       "배경 유리문에 없던 대형 경찰 엠블럼 2개와 영문/한글 문자 밴드가 삽입 — 'No text, captions, watermarks or annotations anywhere' 위반이자 SHOT BACKGROUND를 그대로 유지하라는 락 위반",
       "원본의 투명 유리 자동문이 불투명 흰 패널로 대체되어 로케이션 픽스처가 변형됨"
      ],
      "physics": "A와 동일하게 두 사람은 지면 지지, 여자의 두 손이 남자 팔뚝에 접촉, 당기는 힘의 방향과 옷 주름 일치. 물리적 문제 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "샷 텍스트의 팔뚝 움켜쥔 순간과 좌-여/우-남 중경 미디엄샷을 정확히 구현했고, 배경 경찰서 유리문·볼라드·차량 등 로케이션 락을 원본 그대로 유지했다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "인물 연기·구도는 A와 사실상 동일하나, 아무도 요구하지 않은 배경 개조 — 투명 유리문을 흰 패널로 바꾸고 대형 경찰 엠블럼과 영문 문자 밴드를 삽입 — 로 '텍스트 금지'와 배경 유지 락을 동시에 위반해 얻은 것보다 잃은 것이 크다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "샷 텍스트의 팔뚝 움켜쥔 순간과 좌-여/우-남 중경 미디엄샷을 정확히 구현했고, 배경 경찰서 유리문·볼라드·차량 등 로케이션 락을 원본 그대로 유지했다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "인물 연기·구도는 A와 사실상 동일하나, 아무도 요구하지 않은 배경 개조 — 투명 유리문을 흰 패널로 바꾸고 대형 경찰 엠블럼과 영문 문자 밴드를 삽입 — 로 '텍스트 금지'와 배경 유지 락을 동시에 위반해 얻은 것보다 잃은 것이 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "강민숙은 화면 오른쪽(정인우 쪽)을 향해 시선과 상체를 기울이고 두 손이 그의 오른팔 아래팔을 정면으로 움켜쥐고 있다. 정인우는 오른쪽(프레임 밖 출구 방향)으로 몸을 틀어 나가려는 자세이며 시선도 오른쪽 전방을 향한다. 스케치의 화살표 방향(여자 시선 오른쪽, 남자 이동 오른쪽)과 일치. 그립이 화면 중앙 대각선을 형성해 이탈을 저지하는 것이 읽힌다.",
      "built_space": "배경은 레퍼런스 경찰서 정면 그대로: 청색 캐노피와 금색 경찰 휘장, 유리 자동문, 스테인리스 볼라드 열, 화강암 보도 턱, 좌측 주차 구역의 검정 세단·흰색 SUV·순찰차. 유리문 안 반사와 세로 명패도 원본과 유사하게 유지. 볼라드 개수는 인물에 가려 일부만 보이지만 규모·원근이 자연스럽다.",
      "entities": "강민숙: 40대 후반 한국 여성, 검은 단발, 좁고 단정한 얼굴, 회색 울 코트+검정 터틀넥 — 레퍼런스와 일치(스카프·가방은 프레이밍상 없음, 감점 사유 아님). 정인우: 20대 중반 한국 남성, 검은 머리(비니 없음), 네이비 재킷+초록 스웨터+체크 셔츠 칼라 — 레퍼런스 의상과 일치. 제3의 인물 없음, 텍스트/워터마크 없음.",
      "hard_violations": [],
      "physics": "두 사람 모두 지면에 발을 딛고 있으며(하단 크롭이지만 체중 이동이 전방으로 기울어 있음), 여자는 앞발에 체중을 실어 뒤로 당기는 힘의 자세, 남자는 팔이 뒤로 끌려 어깨가 당겨진 상태. 여자의 두 손이 실제로 남자의 팔뚝을 감싸 접촉면과 옷 주름이 보인다. 공중에 뜬 물체 없음."
     },
     {
      "label": "B",
      "direction": "A와 거의 동일: 여자 시선·상체는 오른쪽 남자 쪽, 손은 그의 아래팔을 잡고, 남자는 오른쪽으로 돌아서며 시선은 오른쪽 전방. 방향성 자체는 문제 없음.",
      "built_space": "배경 유리 자동문 영역이 원본과 달라졌다. 원본의 투명한 유리(내부 반사·안내문·작은 경찰 마크)가 불투명한 흰 패널로 바뀌고, 그 위에 크게 확대된 두 개의 경찰 엠블럼과 판독 불가한 영문/한글 문구('SEOUL METROPOLITAN POLICE AGENCY' 유사 텍스트) 밴드가 문 전체에 붙었다. 즉 배경 픽스처가 변경되고 문자 요소가 추가됨. 볼라드·차량·캐노피는 동일.",
      "entities": "인물 동일: 강민숙(회색 코트, 검은 단발), 정인우(네이비 재킷, 초록 스웨터). 인물 측면에서는 A와 사실상 차이가 없다. 다만 배경에 계약이 금지한 텍스트/캡션성 표기가 다수 삽입되어 있다.",
      "hard_violations": [
       "배경 유리문에 없던 대형 경찰 엠블럼 2개와 영문/한글 문자 밴드가 삽입 — 'No text, captions, watermarks or annotations anywhere' 위반이자 SHOT BACKGROUND를 그대로 유지하라는 락 위반",
       "원본의 투명 유리 자동문이 불투명 흰 패널로 대체되어 로케이션 픽스처가 변형됨"
      ],
      "physics": "A와 동일하게 두 사람은 지면 지지, 여자의 두 손이 남자 팔뚝에 접촉, 당기는 힘의 방향과 옷 주름 일치. 물리적 문제 없음."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "정인우는 프레임 오른쪽에서 오른쪽(출입구 쪽)으로 몸을 돌려 걸어 나가려는 방향이며 시선도 오른쪽 화면 밖을 향한다. 강민숙은 중앙 좌측에서 몸과 시선을 오른쪽(그의 얼굴/등 쪽)으로 향하고 두 손이 그의 왼팔뚝을 잡는 대각선을 만든다. 스케치의 화살표 방향(둘 다 오른쪽)과 일치.",
      "built_space": "경찰서 정면 유리 출입구, 파란 캐노피, 금색 경찰 휘장, 볼라드 열, 흰색 SUV·검정 세단·순찰차, 주차선이 배경 사진과 대응. 다만 유리문 안쪽에 'SEOUL METROPOLITAN POLICE AGENCY' 류의 영문 텍스트 배너와 반복된 휘장 포스터가 새로 크게 삽입되어 있고, 원 배경의 유리 반사/투과 톤이 흐릿한 흰 패널로 바뀌었다. 볼라드 개수도 배경보다 적게 정리됨.",
      "entities": "여성: 40대 후반 한국인, 어깨 길이 검은 단발, 회색 울 코트+검정 터틀넥 — 레퍼런스 일치. 남성: 20대 중반 한국인, 검은 머리(비니 없음), 네이비 재킷+녹색 스웨터+체크 셔츠 — 레퍼런스 의상 일치. 다만 얼굴이 레퍼런스보다 다소 각지고 목이 길게 늘어져 보이며 좌측 팔뚝이 부자연스럽게 길게 뻗어 있다.",
      "hard_violations": [
       "배경 유리문에 원본 배경에 없던 영문 문구/포스터 텍스트가 삽입됨(문구·주석 금지 위반)"
      ],
      "physics": "두 사람 모두 지면에 발을 딛고 있으며(하단 크롭으로 발은 안 보임) 여성은 앞으로 체중을 실어 두 손으로 그의 팔뚝을 잡고, 남성은 팔이 뒤로 당겨져 상체가 멈칫한 자세. 공중에 뜬 물체 없음. 다만 남성의 붙잡힌 팔이 어깨-팔꿈치 비율상 비정상적으로 길게 늘어나 있다."
     },
     {
      "label": "B",
      "direction": "동일하게 정인우는 오른쪽으로 돌아서며 시선은 화면 오른쪽 밖, 강민숙은 오른쪽 그를 향해 몸을 기울이고 두 손으로 그의 왼팔뚝을 붙잡아 중앙 대각선을 형성. 스케치 화살표와 일치.",
      "built_space": "배경 레퍼런스와 거의 동일: 파란 캐노피, 금색 휘장, 유리 자동문과 문 안쪽 원형 경찰 엠블럼(원본 포스터와 대응), 볼라드 열, 흰 SUV·순찰차·검정 세단, 주차선·황색선까지 유지. 유리의 반사·투과 톤도 원본에 가깝다. 인물 뒤 볼라드 수가 배경보다 약간 줄었으나 사소.",
      "entities": "여성: 40대 후반, 검은 단발, 회색 코트+검정 터틀넥 — 레퍼런스 일치. 남성: 20대 중반, 검은 머리, 네이비 재킷+녹색 스웨터+체크 셔츠 — 일치. 얼굴이 레퍼런스의 젊고 준수한 인상에 더 가깝다.",
      "hard_violations": [],
      "physics": "여성은 앞으로 체중을 실은 스탠스로 지면에 서 있고 두 손이 실제로 남성 팔뚝을 감싸 쥔다. 남성은 상체가 오른쪽으로 회전하다 팔이 당겨져 정지된 순간, 접촉점에서 소매 주름이 생김. 부유하는 물체 없음. 팔 길이는 A보다 자연스럽지만 여전히 약간 길다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "구도·동작·의상은 요구대로지만, 배경 유리문에 원본에 없던 영문 문구/포스터가 삽입되어 '텍스트 금지'와 '배경 그대로 유지' 락을 깼고 남성 얼굴·팔 비례도 레퍼런스에서 더 멀어졌다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "미디엄 숏 구도, 좌중앙 강민숙의 다급한 얼굴 위·중앙 대각선 그립, 오른쪽으로 돌아서다 멈춘 정인우가 모두 충족되고 배경 유리문·엠블럼·볼라드·차량이 배경 레퍼런스와 일치하며 삽입 텍스트가 없어 변경이 가장 적다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "구도·동작·의상은 요구대로지만, 배경 유리문에 원본에 없던 영문 문구/포스터가 삽입되어 '텍스트 금지'와 '배경 그대로 유지' 락을 깼고 남성 얼굴·팔 비례도 레퍼런스에서 더 멀어졌다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "미디엄 숏 구도, 좌중앙 강민숙의 다급한 얼굴 위·중앙 대각선 그립, 오른쪽으로 돌아서다 멈춘 정인우가 모두 충족되고 배경 유리문·엠블럼·볼라드·차량이 배경 레퍼런스와 일치하며 삽입 텍스트가 없어 변경이 가장 적다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "정인우는 프레임 오른쪽에서 오른쪽(출입구 쪽)으로 몸을 돌려 걸어 나가려는 방향이며 시선도 오른쪽 화면 밖을 향한다. 강민숙은 중앙 좌측에서 몸과 시선을 오른쪽(그의 얼굴/등 쪽)으로 향하고 두 손이 그의 왼팔뚝을 잡는 대각선을 만든다. 스케치의 화살표 방향(둘 다 오른쪽)과 일치.",
      "built_space": "경찰서 정면 유리 출입구, 파란 캐노피, 금색 경찰 휘장, 볼라드 열, 흰색 SUV·검정 세단·순찰차, 주차선이 배경 사진과 대응. 다만 유리문 안쪽에 'SEOUL METROPOLITAN POLICE AGENCY' 류의 영문 텍스트 배너와 반복된 휘장 포스터가 새로 크게 삽입되어 있고, 원 배경의 유리 반사/투과 톤이 흐릿한 흰 패널로 바뀌었다. 볼라드 개수도 배경보다 적게 정리됨.",
      "entities": "여성: 40대 후반 한국인, 어깨 길이 검은 단발, 회색 울 코트+검정 터틀넥 — 레퍼런스 일치. 남성: 20대 중반 한국인, 검은 머리(비니 없음), 네이비 재킷+녹색 스웨터+체크 셔츠 — 레퍼런스 의상 일치. 다만 얼굴이 레퍼런스보다 다소 각지고 목이 길게 늘어져 보이며 좌측 팔뚝이 부자연스럽게 길게 뻗어 있다.",
      "hard_violations": [
       "배경 유리문에 원본 배경에 없던 영문 문구/포스터 텍스트가 삽입됨(문구·주석 금지 위반)"
      ],
      "physics": "두 사람 모두 지면에 발을 딛고 있으며(하단 크롭으로 발은 안 보임) 여성은 앞으로 체중을 실어 두 손으로 그의 팔뚝을 잡고, 남성은 팔이 뒤로 당겨져 상체가 멈칫한 자세. 공중에 뜬 물체 없음. 다만 남성의 붙잡힌 팔이 어깨-팔꿈치 비율상 비정상적으로 길게 늘어나 있다."
     },
     {
      "label": "A",
      "direction": "동일하게 정인우는 오른쪽으로 돌아서며 시선은 화면 오른쪽 밖, 강민숙은 오른쪽 그를 향해 몸을 기울이고 두 손으로 그의 왼팔뚝을 붙잡아 중앙 대각선을 형성. 스케치 화살표와 일치.",
      "built_space": "배경 레퍼런스와 거의 동일: 파란 캐노피, 금색 휘장, 유리 자동문과 문 안쪽 원형 경찰 엠블럼(원본 포스터와 대응), 볼라드 열, 흰 SUV·순찰차·검정 세단, 주차선·황색선까지 유지. 유리의 반사·투과 톤도 원본에 가깝다. 인물 뒤 볼라드 수가 배경보다 약간 줄었으나 사소.",
      "entities": "여성: 40대 후반, 검은 단발, 회색 코트+검정 터틀넥 — 레퍼런스 일치. 남성: 20대 중반, 검은 머리, 네이비 재킷+녹색 스웨터+체크 셔츠 — 일치. 얼굴이 레퍼런스의 젊고 준수한 인상에 더 가깝다.",
      "hard_violations": [],
      "physics": "여성은 앞으로 체중을 실은 스탠스로 지면에 서 있고 두 손이 실제로 남성 팔뚝을 감싸 쥔다. 남성은 상체가 오른쪽으로 회전하다 팔이 당겨져 정지된 순간, 접촉점에서 소매 주름이 생김. 부유하는 물체 없음. 팔 길이는 A보다 자연스럽지만 여전히 약간 길다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 9
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S7sh5__bgfirst_bg.png",
   "bg_asset_id": "9b5b40a9-478e-4d43-99a4-abf9dac90c8b",
   "bg_record_key": "S7sh5::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S9sh2::bgfirst_bg": {
  "input_fingerprint": "6143892570558c73",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 거실 소파에 등을 보이고 앉은 강영란(흰 머리 여자)의 하얗게 샌 머리가 가지런히 묶인 뒷모습.\n\nLOCATION (lock): The modest living room of a small two-room multi-family home, centered on its sofa near the entrance.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track into the prescribed eye-level position behind 강영란(흰 머리 여자), then settle into a tight close view centered on her neatly tied white hair and upper back. The sofa back forms a low horizontal boundary beneath her shoulders, while her head remains turned far enough away that no facial contour becomes visible; the emphasized change is the tighter camera distance.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: living-room sofa (occupied by the seated woman) — The rear and upper edge of the sofa face the camera; used as Its backrest forms the lower framing line beneath the seated woman's shoulders.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient illumination appropriate to the nighttime interior separates the white hair from the darker surrounding values.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 거실 소파에 등을 보이고 앉은 강영란(흰 머리 여자)의 하얗게 샌 머리가 가지런히 묶인 뒷모습.\n\nLOCATION (lock): The modest living room of a small two-room multi-family home, centered on its sofa near the entrance.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track into the prescribed eye-level position behind 강영란(흰 머리 여자), then settle into a tight close view centered on her neatly tied white hair and upper back. The sofa back forms a low horizontal boundary beneath her shoulders, while her head remains turned far enough away that no facial contour becomes visible; the emphasized change is the tighter camera distance.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: living-room sofa (occupied by the seated woman) — The rear and upper edge of the sofa face the camera; used as Its backrest forms the lower framing line beneath the seated woman's shoulders.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient illumination appropriate to the nighttime interior separates the white hair from the darker surrounding values.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S9sh2__bgfirst_bg.png",
  "asset_id": "3434379a-7ac6-4667-9244-dc3f8b2bd2ca",
  "input_asset_ids": [
   "175b12ef-2596-497b-9fa8-37c95c2b3459",
   "41400a21-5b9d-4d6d-bfa4-ebb3bff48a6f"
  ]
 },
 "S9sh2": {
  "input_fingerprint": "d49473a22f8b045d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 거실 소파에 등을 보이고 앉은 강영란(흰 머리 여자)의 하얗게 샌 머리가 가지런히 묶인 뒷모습.\n\nLOCATION (lock): The modest living room of a small two-room multi-family home, centered on its sofa near the entrance. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track into the prescribed eye-level position behind 강영란(흰 머리 여자), then settle into a tight close view centered on her neatly tied white hair and upper back. The sofa back forms a low horizontal boundary beneath her shoulders, while her head remains turned far enough away that no facial contour becomes visible; the emphasized change is the tighter camera distance.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: living-room sofa (occupied by the seated woman) — The rear and upper edge of the sofa face the camera; used as Its backrest forms the lower framing line beneath the seated woman's shoulders.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient illumination appropriate to the nighttime interior separates the white hair from the darker surrounding values.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (Korean woman in her late 60s, neatly twisted white hair, graceful dignified face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 거실 소파에 등을 보이고 앉은 강영란(흰 머리 여자)의 하얗게 샌 머리가 가지런히 묶인 뒷모습.\n\nLOCATION (lock): The modest living room of a small two-room multi-family home, centered on its sofa near the entrance. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track into the prescribed eye-level position behind 강영란(흰 머리 여자), then settle into a tight close view centered on her neatly tied white hair and upper back. The sofa back forms a low horizontal boundary beneath her shoulders, while her head remains turned far enough away that no facial contour becomes visible; the emphasized change is the tighter camera distance.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: living-room sofa (occupied by the seated woman) — The rear and upper edge of the sofa face the camera; used as Its backrest forms the lower framing line beneath the seated woman's shoulders.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient illumination appropriate to the nighttime interior separates the white hair from the darker surrounding values.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (Korean woman in her late 60s, neatly twisted white hair, graceful dignified face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 거실 소파에 등을 보이고 앉은 강영란(흰 머리 여자)의 하얗게 샌 머리가 가지런히 묶인 뒷모습.\n\nLOCATION (lock): The modest living room of a small two-room multi-family home, centered on its sofa near the entrance. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track into the prescribed eye-level position behind 강영란(흰 머리 여자), then settle into a tight close view centered on her neatly tied white hair and upper back. The sofa back forms a low horizontal boundary beneath her shoulders, while her head remains turned far enough away that no facial contour becomes visible; the emphasized change is the tighter camera distance.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: living-room sofa (occupied by the seated woman) — The rear and upper edge of the sofa face the camera; used as Its backrest forms the lower framing line beneath the seated woman's shoulders.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient illumination appropriate to the nighttime interior separates the white hair from the darker surrounding values.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (Korean woman in her late 60s, neatly twisted white hair, graceful dignified face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S9sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S9sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1013203>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L08B02.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1013203>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "장소·정체성·야간 조도는 잘 맞으나 요구된 '타이트한 클로즈업'보다 훨씬 넓은 미디엄 프레임이라 머리·상체가 작게 잡히고, 소파 등받이가 어깨 아래 낮은 경계선으로 기능하지 못한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "묶은 백발과 상부 등을 화면 중앙에 크게 잡은 타이트 클로즈업, 소파 등받이가 어깨 아래 수평 경계선을 이루는 구성까지 샷 텍스트를 정확히 실현했고 로케이션 요소도 일치한다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.528,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.528,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "장소·정체성·야간 조도는 잘 맞으나 요구된 '타이트한 클로즈업'보다 훨씬 넓은 미디엄 프레임이라 머리·상체가 작게 잡히고, 소파 등받이가 어깨 아래 낮은 경계선으로 기능하지 못한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "묶은 백발과 상부 등을 화면 중앙에 크게 잡은 타이트 클로즈업, 소파 등받이가 어깨 아래 수평 경계선을 이루는 구성까지 샷 텍스트를 정확히 실현했고 로케이션 요소도 일치한다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 소파 뒤 눈높이에서 앉은 여성의 뒤통수를 정면으로 바라본다. 인물의 시선은 카메라 반대쪽(거실 안쪽/창 방향)으로 완전히 향해 있어 얼굴 윤곽이 전혀 보이지 않는다. 머리 후두부의 낮게 묶은 시뇽이 화면 중앙에 정확히 위치한다.",
      "built_space": "레퍼런스 거실의 요소가 모두 확인된다 — 좌측 발코니 슬라이딩 창(야경), 좌측 회색 패브릭 소파와 짙은 쿠션, 창 옆 원형 구멍 뚫린 세로 패널(스피커), 우측 현관 복도·디지털 도어록 달린 흰 현관문·타일 바닥·붙박이장. 소파 등받이 상단이 프레임 하단을 가로지르는 낮은 수평선으로 어깨 아래를 자른다. 소파는 하나, 인물은 한 명, 중복 구조물 없음.",
      "entities": "인물은 백발을 가지런히 꼬아 묶은 60대 후반 한국 여성으로 캐릭터 레퍼런스의 헤어 스타일·체구와 부합. 얼굴은 규정대로 비노출. 어두운 자주빛 상의. 추가 인물 없음. 텍스트·워터마크 없음.",
      "hard_violations": [],
      "physics": "인물은 소파 좌석에 앉아 등받이에 등을 기대고 있으며, 어깨선 아래는 등받이에 가려진다. 공중에 뜬 신체나 무지지 물체 없음. 머리카락 결과 잔머리가 중력에 맞게 정리되어 있고, 패브릭 소파의 직조 질감·주름이 실물감 있게 표현됨."
     },
     {
      "label": "B",
      "direction": "카메라는 소파 뒤 눈높이에서 인물의 뒤통수와 등을 본다. 인물은 카메라 반대편(현관/벽 방향)을 향하고 얼굴은 보이지 않는다. 다만 인물이 화면 중앙보다 다소 작게 잡혀 미디엄 숏에 가깝다.",
      "built_space": "레퍼런스 요소 — 좌측 슬라이딩 창, 원형 구멍 패널, 우측 현관 복도·도어록 문·타일·붙박이장 — 모두 존재. 소파 등받이가 프레임 하단을 가로지르나 인물의 어깨보다 훨씬 아래에 있어 상체 대부분과 등판 전체가 노출된다. 인물이 소파 좌석면보다 앞쪽/아래에 앉은 듯 등받이와 몸의 접점이 애매하다.",
      "entities": "백발 쪽머리(낮은 번)의 노년 한국 여성, 회갈색 셔츠. 캐릭터 레퍼런스와 대체로 일치. 다른 인물이나 삽입 오브젝트 없음. 텍스트 없음.",
      "hard_violations": [],
      "physics": "앉은 자세는 성립하나 소파 등받이 위로 상체가 상당히 떠 보여 좌석 위치가 불명확하다. 공중 부양 등 명백한 무지지 요소는 없음. 재질감은 전반적으로 평이하고 머리 디테일이 다소 뭉개져 있다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 9,
      "verdict_ko": "지정된 '클로즈업' 스케일과 '소파 등받이가 어깨 아래 낮은 수평 경계' 조건을 정확히 충족하고, 백발 쪽머리가 중앙에 크게 잡히며 얼굴 비노출·야간 저조도 조건도 맞아 최상."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "장소 재현과 뒷모습 방향은 맞지만 프레임이 넓어 클로즈업이 아닌 미디엄 숏이 되었고, 등받이가 어깨 훨씬 아래에 있어 '어깨 아래 낮은 경계' 지시와 어긋난다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "지정된 '클로즈업' 스케일과 '소파 등받이가 어깨 아래 낮은 수평 경계' 조건을 정확히 충족하고, 백발 쪽머리가 중앙에 크게 잡히며 얼굴 비노출·야간 저조도 조건도 맞아 최상."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "장소 재현과 뒷모습 방향은 맞지만 프레임이 넓어 클로즈업이 아닌 미디엄 숏이 되었고, 등받이가 어깨 훨씬 아래에 있어 '어깨 아래 낮은 경계' 지시와 어긋난다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "카메라는 소파 뒤 눈높이에서 앉은 여성의 뒤통수를 정면으로 바라본다. 인물의 시선은 카메라 반대쪽(거실 안쪽/창 방향)으로 완전히 향해 있어 얼굴 윤곽이 전혀 보이지 않는다. 머리 후두부의 낮게 묶은 시뇽이 화면 중앙에 정확히 위치한다.",
      "built_space": "레퍼런스 거실의 요소가 모두 확인된다 — 좌측 발코니 슬라이딩 창(야경), 좌측 회색 패브릭 소파와 짙은 쿠션, 창 옆 원형 구멍 뚫린 세로 패널(스피커), 우측 현관 복도·디지털 도어록 달린 흰 현관문·타일 바닥·붙박이장. 소파 등받이 상단이 프레임 하단을 가로지르는 낮은 수평선으로 어깨 아래를 자른다. 소파는 하나, 인물은 한 명, 중복 구조물 없음.",
      "entities": "인물은 백발을 가지런히 꼬아 묶은 60대 후반 한국 여성으로 캐릭터 레퍼런스의 헤어 스타일·체구와 부합. 얼굴은 규정대로 비노출. 어두운 자주빛 상의. 추가 인물 없음. 텍스트·워터마크 없음.",
      "hard_violations": [],
      "physics": "인물은 소파 좌석에 앉아 등받이에 등을 기대고 있으며, 어깨선 아래는 등받이에 가려진다. 공중에 뜬 신체나 무지지 물체 없음. 머리카락 결과 잔머리가 중력에 맞게 정리되어 있고, 패브릭 소파의 직조 질감·주름이 실물감 있게 표현됨."
     },
     {
      "label": "A",
      "direction": "카메라는 소파 뒤 눈높이에서 인물의 뒤통수와 등을 본다. 인물은 카메라 반대편(현관/벽 방향)을 향하고 얼굴은 보이지 않는다. 다만 인물이 화면 중앙보다 다소 작게 잡혀 미디엄 숏에 가깝다.",
      "built_space": "레퍼런스 요소 — 좌측 슬라이딩 창, 원형 구멍 패널, 우측 현관 복도·도어록 문·타일·붙박이장 — 모두 존재. 소파 등받이가 프레임 하단을 가로지르나 인물의 어깨보다 훨씬 아래에 있어 상체 대부분과 등판 전체가 노출된다. 인물이 소파 좌석면보다 앞쪽/아래에 앉은 듯 등받이와 몸의 접점이 애매하다.",
      "entities": "백발 쪽머리(낮은 번)의 노년 한국 여성, 회갈색 셔츠. 캐릭터 레퍼런스와 대체로 일치. 다른 인물이나 삽입 오브젝트 없음. 텍스트 없음.",
      "hard_violations": [],
      "physics": "앉은 자세는 성립하나 소파 등받이 위로 상체가 상당히 떠 보여 좌석 위치가 불명확하다. 공중 부양 등 명백한 무지지 요소는 없음. 재질감은 전반적으로 평이하고 머리 디테일이 다소 뭉개져 있다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 17
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 12,
   "B": 17
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "장소·정체성·야간 조도는 잘 맞으나 요구된 '타이트한 클로즈업'보다 훨씬 넓은 미디엄 프레임이라 머리·상체가 작게 잡히고, 소파 등받이가 어깨 아래 낮은 경계선으로 기능하지 못한다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "묶은 백발과 상부 등을 화면 중앙에 크게 잡은 타이트 클로즈업, 소파 등받이가 어깨 아래 수평 경계선을 이루는 구성까지 샷 텍스트를 정확히 실현했고 로케이션 요소도 일치한다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L08B02.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1013203>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "여자가 거실 소파에 앉아 있어야 하나, 화면 전경에 임의의 소파 등받이가 추가되었고 배경 왼쪽에는 실제 거실 소파가 빈 채로 중복 배치되어 공간 구조가 어긋났습니다.",
     "fix_en": "Remove the unoccupied sofa from the background left, filling that area with the continuous wooden floor, lower wall, and lower window frame, so the foreground sofa the woman sits on reads as the only one in the room. Preserve the woman's figure, the foreground sofa backrest, and the rest of the room's layout and lighting exactly.",
     "needs_regeneration": true
    },
    {
     "issue_ko": "여자가 입은 어두운 단색 상의는 캐릭터 레퍼런스 이미지에 지정된 화려한 금박 무늬의 붉은색 전통 한복 의상과 일치하지 않습니다.",
     "fix_en": "Change the woman's dark top into the back view of the bright red traditional hanbok top with intricate gold stamped patterns and a white collar, matching the character reference. Preserve her white tied hair, her posture, the foreground sofa backrest, and the entire background room geometry and lighting.",
     "needs_regeneration": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the unoccupied sofa from the background left, filling that area with the continuous wooden floor, lower wall, and lower window frame, so the foreground sofa the woman sits on reads as the only one in the room. Preserve the woman's figure, the foreground sofa backrest, and the rest of the room's layout and lighting exactly.\n- Change the woman's dark top into the back view of the bright red traditional hanbok top with intricate gold stamped patterns and a white collar, matching the character reference. Preserve her white tied hair, her posture, the foreground sofa backrest, and the entire background room geometry and lighting.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 소파 뒤 눈높이에서 여성의 뒤통수를 정면으로 바라본다. 얼굴 윤곽은 전혀 보이지 않고 시선 방향은 화면 안쪽(창/현관 반대편)으로 향한다. 인물은 중앙 중단에 배치되고, 하단에 소파 등받이 상단이 가로 프레임 라인을 이룬다. 샷 텍스트의 '등을 보이고 앉은 뒷모습'과 일치.",
      "built_space": "레퍼런스 거실과 일치: 좌측 발코니 슬라이딩 창(야경), 우측 현관 복도와 도어락 달린 현대식 문, 천장 매입등, 붙박이장, 회색 패브릭 소파(등받이가 하단 프레임), 소파 위 쿠션, 벽에 기댄 스피커 패널까지 보인다. 소파는 하나, 인물은 소파에 앉아 있어 좌석 점유가 올바르다.",
      "entities": "백발을 단정히 낮게 트위스트로 묶은 노년 한국 여성의 뒷머리와 어깨 상부. 머리 결·잔머리·두피가 실제 모발로 읽힌다. 의상은 어두운 자주/검정 계열 현대 상의로, 현대 한국 밤 거실 설정에 맞는다. 다른 인물 없음.",
      "hard_violations": [],
      "physics": "인물은 소파 좌석에 앉아 등받이 뒤로 상체가 솟아 있으며, 어깨선이 등받이 위쪽에 자연스럽게 걸린다. 공중에 뜬 것 없음. 머리 무게와 목의 각도가 자연스럽다."
     },
     {
      "label": "B",
      "direction": "역시 뒤통수 정면 뷰이며 얼굴은 보이지 않는다. 다만 프레임이 훨씬 넓어 클로즈업이 아니라 미디엄/와이드에 가깝고, 인물의 등 전체와 방 전체가 들어온다. 하단의 소파 등받이는 인물의 어깨가 아니라 화면 최하단 별도의 덩어리로 놓여 인물과 분리되어 보인다.",
      "built_space": "레퍼런스 거실 요소(창, 현관, 붙박이장, 스피커 패널, 콘센트)는 잘 재현되었으나 조명이 야간 저조도가 아니라 균일하고 밝은 편이다. 결정적으로 인물이 앉아 있는 좌석이 보이지 않는다 — 화면 하단 소파는 인물보다 훨씬 앞(카메라 쪽)에 있고, 인물의 몸통은 소파 좌석면이 아닌 마루 위 허공/불명확한 위치에서 시작한다. 소파에 앉은 것으로 읽히지 않는다.",
      "entities": "백발을 낮게 묶은 노년 여성이지만, 의상이 캐릭터 레퍼런스의 붉은 활옷(금색 자수)을 그대로 입고 있어 현대 한국 밤 거실 설정과 충돌한다. 계약은 의상을 지정하지 않았고 레퍼런스는 정체성(얼굴·머리·체구)용이므로 혼례복은 요청되지 않은 변경이다.",
      "hard_violations": [
       "인물이 소파에 앉아 있다는 스테이징이 성립하지 않음 — 하단 소파는 별개의 앞쪽 덩어리이고 인물의 하체를 지지하는 좌석이 보이지 않아 허공/바닥에 떠 있는 배치로 읽힘"
      ],
      "physics": "인물 상체 아래에 지지 구조가 확인되지 않는다. 화면 하단의 소파 등받이는 인물보다 수 미터 앞에 있어 그 소파에 앉은 것이 될 수 없고, 인물 뒤쪽 바닥은 그대로 마루라 앉을 가구가 없다. 즉 지지물이 보이지 않는다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "소파 등받이가 어깨 아래 수평 경계를 이루는 눈높이 타이트 클로즈업, 야간 저조도에서 백발만 분리되는 톤, 현대 거실 로케이션 일치 — 샷 텍스트의 프레이밍·순간을 정확히 실현했다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "얼굴 비노출은 지켰으나 클로즈업이 아닌 광각으로 넓혔고, 조명이 야간 저조도가 아니며, 요청되지 않은 붉은 혼례복을 입혔고, 무엇보다 인물이 앉은 소파 좌석이 없어 '소파에 앉은' 스테이징이 물리적으로 성립하지 않는다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "소파 등받이가 어깨 아래 수평 경계를 이루는 눈높이 타이트 클로즈업, 야간 저조도에서 백발만 분리되는 톤, 현대 거실 로케이션 일치 — 샷 텍스트의 프레이밍·순간을 정확히 실현했다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "얼굴 비노출은 지켰으나 클로즈업이 아닌 광각으로 넓혔고, 조명이 야간 저조도가 아니며, 요청되지 않은 붉은 혼례복을 입혔고, 무엇보다 인물이 앉은 소파 좌석이 없어 '소파에 앉은' 스테이징이 물리적으로 성립하지 않는다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 소파 뒤 눈높이에서 여성의 뒤통수를 정면으로 바라본다. 얼굴 윤곽은 전혀 보이지 않고 시선 방향은 화면 안쪽(창/현관 반대편)으로 향한다. 인물은 중앙 중단에 배치되고, 하단에 소파 등받이 상단이 가로 프레임 라인을 이룬다. 샷 텍스트의 '등을 보이고 앉은 뒷모습'과 일치.",
      "built_space": "레퍼런스 거실과 일치: 좌측 발코니 슬라이딩 창(야경), 우측 현관 복도와 도어락 달린 현대식 문, 천장 매입등, 붙박이장, 회색 패브릭 소파(등받이가 하단 프레임), 소파 위 쿠션, 벽에 기댄 스피커 패널까지 보인다. 소파는 하나, 인물은 소파에 앉아 있어 좌석 점유가 올바르다.",
      "entities": "백발을 단정히 낮게 트위스트로 묶은 노년 한국 여성의 뒷머리와 어깨 상부. 머리 결·잔머리·두피가 실제 모발로 읽힌다. 의상은 어두운 자주/검정 계열 현대 상의로, 현대 한국 밤 거실 설정에 맞는다. 다른 인물 없음.",
      "hard_violations": [],
      "physics": "인물은 소파 좌석에 앉아 등받이 뒤로 상체가 솟아 있으며, 어깨선이 등받이 위쪽에 자연스럽게 걸린다. 공중에 뜬 것 없음. 머리 무게와 목의 각도가 자연스럽다."
     },
     {
      "label": "B",
      "direction": "역시 뒤통수 정면 뷰이며 얼굴은 보이지 않는다. 다만 프레임이 훨씬 넓어 클로즈업이 아니라 미디엄/와이드에 가깝고, 인물의 등 전체와 방 전체가 들어온다. 하단의 소파 등받이는 인물의 어깨가 아니라 화면 최하단 별도의 덩어리로 놓여 인물과 분리되어 보인다.",
      "built_space": "레퍼런스 거실 요소(창, 현관, 붙박이장, 스피커 패널, 콘센트)는 잘 재현되었으나 조명이 야간 저조도가 아니라 균일하고 밝은 편이다. 결정적으로 인물이 앉아 있는 좌석이 보이지 않는다 — 화면 하단 소파는 인물보다 훨씬 앞(카메라 쪽)에 있고, 인물의 몸통은 소파 좌석면이 아닌 마루 위 허공/불명확한 위치에서 시작한다. 소파에 앉은 것으로 읽히지 않는다.",
      "entities": "백발을 낮게 묶은 노년 여성이지만, 의상이 캐릭터 레퍼런스의 붉은 활옷(금색 자수)을 그대로 입고 있어 현대 한국 밤 거실 설정과 충돌한다. 계약은 의상을 지정하지 않았고 레퍼런스는 정체성(얼굴·머리·체구)용이므로 혼례복은 요청되지 않은 변경이다.",
      "hard_violations": [
       "인물이 소파에 앉아 있다는 스테이징이 성립하지 않음 — 하단 소파는 별개의 앞쪽 덩어리이고 인물의 하체를 지지하는 좌석이 보이지 않아 허공/바닥에 떠 있는 배치로 읽힘"
      ],
      "physics": "인물 상체 아래에 지지 구조가 확인되지 않는다. 화면 하단의 소파 등받이는 인물보다 수 미터 앞에 있어 그 소파에 앉은 것이 될 수 없고, 인물 뒤쪽 바닥은 그대로 마루라 앉을 가구가 없다. 즉 지지물이 보이지 않는다."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 소파 뒤 눈높이에서 여자의 뒤통수를 정면으로 본다. 인물은 카메라 반대쪽(현관/벽 쪽)을 향하고 있어 얼굴 윤곽은 전혀 보이지 않는다. 시선 방향은 정면 벽 쪽으로 읽힌다. 다만 카메라 거리가 멀어 '타이트한 클로즈업'이 아니라 미디엄 숏에 가깝고, 방 전체(창, 복도, 현관문, 붙박이장, 스피커, 콘센트)가 넓게 보인다.",
      "built_space": "로케이션 사진의 요소(발코니 새시, 오른쪽 복도와 현관문·도어록, 붙박이장, 다운라이트, 삼각 스피커 패널, 마루)가 거의 그대로 보인다. 그러나 소파는 프레임 하단 가장자리에만 회색 등받이 한 줄로 걸쳐 있고, 인물의 몸통이 소파 등받이보다 훨씬 위·뒤에 떠 있어 인물이 소파에 앉아 있는 것으로 읽히기 어렵다. 어깨 아래 등받이가 프레임 하단선을 이루라는 지시는 형식상 충족되나, 인물과 소파의 거리·스케일이 어긋나 인물이 소파 뒤 바닥에 서 있거나 별도 위치에 있는 것처럼 보인다. 방의 밝기는 야간 저조도라기보다 균일하게 밝은 실내로, 라이팅 락과 어긋난다.",
      "entities": "60대 후반 한국 여성, 백발을 뒤로 단정히 틀어 올린 쪽머리 — 캐릭터 프로필과 일치. 다만 의상이 캐릭터 레퍼런스의 혼례복(활옷) 붉은 금박 상의를 그대로 입고 있어, 현대 한국 거실 야간 장면에는 어울리지 않는 시대·상황 착오적 복장이다. 다른 인물이나 추가 인체는 없고 텍스트·워터마크도 없다.",
      "hard_violations": [
       "소파에 앉은 인물이라는 스테이징이 성립하지 않음: 소파 등받이가 프레임 최하단에 있고 인물 상체는 그보다 한참 뒤·위에 위치해 실제로 그 소파에 앉아 있을 수 없는 공간 배치(인물이 앉은 자리를 스테이징이 제공하지 않음)"
      ],
      "physics": "인물의 상체를 지지하는 것은 화면상 보이지 않는다. 프레임 하단의 소파 등받이는 인물과 상당한 거리·높이 차가 있어 그 소파에 앉은 몸으로 연결되지 않으며, 인물 하단은 소파 쿠션 선에서 잘려 마치 바닥 위에 떠 있거나 별도의 낮은 좌석에 앉은 듯 읽힌다. 옷감은 실크 광택이 있으나 드레이프가 다소 평면적이다."
     },
     {
      "label": "B",
      "direction": "카메라는 소파 등받이 바로 뒤 눈높이에서 여자의 뒤통수를 타이트하게 잡는다. 인물은 카메라 반대쪽, 즉 창/복도 쪽 정면을 향하고 있어 얼굴 윤곽이 전혀 보이지 않는다. 목덜미와 귀 뒤 일부만 보인다.",
      "built_space": "로케이션 요소가 유지된다: 왼쪽 발코니 새시와 야경, 회색 소파 등받이(왼쪽으로 이어짐), 쿠션, 오른쪽 복도·현관문·도어록·붙박이장·다운라이트, 마루. 프레임 하단 전체를 소파 등받이가 가로로 가르며 어깨 아래 하단 프레임선을 정확히 형성한다. 인물은 소파 좌면에 앉아 등받이 위로 머리와 어깨 윗부분만 나온 관계가 기하학적으로 성립한다. 야간 저조도로 백발만 밝게 분리된다.",
      "entities": "백발을 단정히 꼬아 올린 쪽머리의 노년 여성 뒷모습 — 프로필 부합. 목덜미 피부 질감과 잔머리가 노년 여성으로 읽힌다. 의상은 어두운 자주/검정 계열의 단정한 상의로 현대 거실 야간에 자연스럽다(레퍼런스 혼례복은 아니지만 프롬프트가 의상을 지정하지 않음). 추가 인물·텍스트 없음.",
      "hard_violations": [],
      "physics": "인물은 소파 좌면에 앉아 있고, 등받이가 어깨 아래를 가리는 관계로 체중 지지가 명확하다. 머리·어깨 실루엣과 등받이 곡선이 연속적으로 맞물려 떠 있는 부분이 없다. 소파 패브릭의 직조 질감과 압축감, 머리카락 가닥의 광택 모두 물리적으로 읽힌다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "얼굴 비노출은 지켰으나 '타이트한 클로즈업' 대신 방 전체가 보이는 미디엄 숏이고, 소파 등받이가 화면 맨 아래 멀리 떨어져 있어 인물이 그 소파에 앉아 있다는 기하가 성립하지 않으며, 야간 저조도 대신 균일하게 밝고 혼례복 활옷이라는 상황 착오 의상까지 더해 계약 위반이 크다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "소파 등받이가 어깨 아래 수평 하단선을 이루는 타이트 클로즈업, 얼굴 윤곽 완전 비노출, 야간 저조도에서 백발만 분리되는 조명까지 샷 텍스트의 카메라·프레임·라이팅 지시를 모두 충족하고 로케이션 고정 요소(현관문·도어록·붙박이장·발코니 새시)도 유지했다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "얼굴 비노출은 지켰으나 '타이트한 클로즈업' 대신 방 전체가 보이는 미디엄 숏이고, 소파 등받이가 화면 맨 아래 멀리 떨어져 있어 인물이 그 소파에 앉아 있다는 기하가 성립하지 않으며, 야간 저조도 대신 균일하게 밝고 혼례복 활옷이라는 상황 착오 의상까지 더해 계약 위반이 크다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "소파 등받이가 어깨 아래 수평 하단선을 이루는 타이트 클로즈업, 얼굴 윤곽 완전 비노출, 야간 저조도에서 백발만 분리되는 조명까지 샷 텍스트의 카메라·프레임·라이팅 지시를 모두 충족하고 로케이션 고정 요소(현관문·도어록·붙박이장·발코니 새시)도 유지했다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "카메라는 소파 뒤 눈높이에서 여자의 뒤통수를 정면으로 본다. 인물은 카메라 반대쪽(현관/벽 쪽)을 향하고 있어 얼굴 윤곽은 전혀 보이지 않는다. 시선 방향은 정면 벽 쪽으로 읽힌다. 다만 카메라 거리가 멀어 '타이트한 클로즈업'이 아니라 미디엄 숏에 가깝고, 방 전체(창, 복도, 현관문, 붙박이장, 스피커, 콘센트)가 넓게 보인다.",
      "built_space": "로케이션 사진의 요소(발코니 새시, 오른쪽 복도와 현관문·도어록, 붙박이장, 다운라이트, 삼각 스피커 패널, 마루)가 거의 그대로 보인다. 그러나 소파는 프레임 하단 가장자리에만 회색 등받이 한 줄로 걸쳐 있고, 인물의 몸통이 소파 등받이보다 훨씬 위·뒤에 떠 있어 인물이 소파에 앉아 있는 것으로 읽히기 어렵다. 어깨 아래 등받이가 프레임 하단선을 이루라는 지시는 형식상 충족되나, 인물과 소파의 거리·스케일이 어긋나 인물이 소파 뒤 바닥에 서 있거나 별도 위치에 있는 것처럼 보인다. 방의 밝기는 야간 저조도라기보다 균일하게 밝은 실내로, 라이팅 락과 어긋난다.",
      "entities": "60대 후반 한국 여성, 백발을 뒤로 단정히 틀어 올린 쪽머리 — 캐릭터 프로필과 일치. 다만 의상이 캐릭터 레퍼런스의 혼례복(활옷) 붉은 금박 상의를 그대로 입고 있어, 현대 한국 거실 야간 장면에는 어울리지 않는 시대·상황 착오적 복장이다. 다른 인물이나 추가 인체는 없고 텍스트·워터마크도 없다.",
      "hard_violations": [
       "소파에 앉은 인물이라는 스테이징이 성립하지 않음: 소파 등받이가 프레임 최하단에 있고 인물 상체는 그보다 한참 뒤·위에 위치해 실제로 그 소파에 앉아 있을 수 없는 공간 배치(인물이 앉은 자리를 스테이징이 제공하지 않음)"
      ],
      "physics": "인물의 상체를 지지하는 것은 화면상 보이지 않는다. 프레임 하단의 소파 등받이는 인물과 상당한 거리·높이 차가 있어 그 소파에 앉은 몸으로 연결되지 않으며, 인물 하단은 소파 쿠션 선에서 잘려 마치 바닥 위에 떠 있거나 별도의 낮은 좌석에 앉은 듯 읽힌다. 옷감은 실크 광택이 있으나 드레이프가 다소 평면적이다."
     },
     {
      "label": "A",
      "direction": "카메라는 소파 등받이 바로 뒤 눈높이에서 여자의 뒤통수를 타이트하게 잡는다. 인물은 카메라 반대쪽, 즉 창/복도 쪽 정면을 향하고 있어 얼굴 윤곽이 전혀 보이지 않는다. 목덜미와 귀 뒤 일부만 보인다.",
      "built_space": "로케이션 요소가 유지된다: 왼쪽 발코니 새시와 야경, 회색 소파 등받이(왼쪽으로 이어짐), 쿠션, 오른쪽 복도·현관문·도어록·붙박이장·다운라이트, 마루. 프레임 하단 전체를 소파 등받이가 가로로 가르며 어깨 아래 하단 프레임선을 정확히 형성한다. 인물은 소파 좌면에 앉아 등받이 위로 머리와 어깨 윗부분만 나온 관계가 기하학적으로 성립한다. 야간 저조도로 백발만 밝게 분리된다.",
      "entities": "백발을 단정히 꼬아 올린 쪽머리의 노년 여성 뒷모습 — 프로필 부합. 목덜미 피부 질감과 잔머리가 노년 여성으로 읽힌다. 의상은 어두운 자주/검정 계열의 단정한 상의로 현대 거실 야간에 자연스럽다(레퍼런스 혼례복은 아니지만 프롬프트가 의상을 지정하지 않음). 추가 인물·텍스트 없음.",
      "hard_violations": [],
      "physics": "인물은 소파 좌면에 앉아 있고, 등받이가 어깨 아래를 가리는 관계로 체중 지지가 명확하다. 머리·어깨 실루엣과 등받이 곡선이 연속적으로 맞물려 떠 있는 부분이 없다. 소파 패브릭의 직조 질감과 압축감, 머리카락 가닥의 광택 모두 물리적으로 읽힌다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 6
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S9sh2__bgfirst_bg.png",
   "bg_asset_id": "3434379a-7ac6-4667-9244-dc3f8b2bd2ca",
   "bg_record_key": "S9sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S10sh1::bgfirst_bg": {
  "input_fingerprint": "52dece4020edd09e",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 편의점 카운터에 서서 두툼한 돈봉투를 쥔 채 멍한 눈으로 허공을 응시하는 김지후(강수리영)의 모습.\n\nLOCATION (lock): Behind the convenience-store checkout counter on the sales floor at night.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Observe 김지후(강수리영) frontally from the prescribed eye-level position across the counter, using a slightly wide medium frame that includes her upper body and both hands around the thick money envelope. She sits near frame center behind the counter with vacant eyes fixed beyond the lens, while the envelope remains clearly readable in the lower middle; the emphasized axis is her static, isolated position.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: convenience-store counter (김지후 positioned behind it) — The customer-facing side and top edge are visible across the lower frame; used as Creates the horizontal foreground boundary separating the observer from 김지후; thick money envelope (held and being handled) — Its broad side is angled partly toward the camera between her hands; used as Held within the lower central portion of the composition as the physical focus of her restless hands.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Ambient convenience-store illumination renders the present space clearly while the restrained, high-contrast palette keeps her expression emotionally remote.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 편의점 카운터에 서서 두툼한 돈봉투를 쥔 채 멍한 눈으로 허공을 응시하는 김지후(강수리영)의 모습.\n\nLOCATION (lock): Behind the convenience-store checkout counter on the sales floor at night.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Observe 김지후(강수리영) frontally from the prescribed eye-level position across the counter, using a slightly wide medium frame that includes her upper body and both hands around the thick money envelope. She sits near frame center behind the counter with vacant eyes fixed beyond the lens, while the envelope remains clearly readable in the lower middle; the emphasized axis is her static, isolated position.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: convenience-store counter (김지후 positioned behind it) — The customer-facing side and top edge are visible across the lower frame; used as Creates the horizontal foreground boundary separating the observer from 김지후; thick money envelope (held and being handled) — Its broad side is angled partly toward the camera between her hands; used as Held within the lower central portion of the composition as the physical focus of her restless hands.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Ambient convenience-store illumination renders the present space clearly while the restrained, high-contrast palette keeps her expression emotionally remote.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S10sh1__bgfirst_bg.png",
  "asset_id": "aded442b-127e-4feb-8b3f-28bbaf0e4913",
  "input_asset_ids": [
   "04f58860-ddd4-4df1-8170-329d5e1544e3",
   "303651ce-0bb1-462b-900d-4f035be71d92"
  ]
 },
 "S10sh1": {
  "input_fingerprint": "5ea15eac1616510a",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 편의점 카운터에 서서 두툼한 돈봉투를 쥔 채 멍한 눈으로 허공을 응시하는 김지후(강수리영)의 모습.\n\nLOCATION (lock): Behind the convenience-store checkout counter on the sales floor at night. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Observe 김지후(강수리영) frontally from the prescribed eye-level position across the counter, using a slightly wide medium frame that includes her upper body and both hands around the thick money envelope. She sits near frame center behind the counter with vacant eyes fixed beyond the lens, while the envelope remains clearly readable in the lower middle; the emphasized axis is her static, isolated position.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: convenience-store counter (김지후 positioned behind it) — The customer-facing side and top edge are visible across the lower frame; used as Creates the horizontal foreground boundary separating the observer from 김지후; thick money envelope (held and being handled) — Its broad side is angled partly toward the camera between her hands; used as Held within the lower central portion of the composition as the physical focus of her restless hands.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Ambient convenience-store illumination renders the present space clearly while the restrained, high-contrast palette keeps her expression emotionally remote.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her white animal-bone necklace and holds the thick cash envelope at the counter.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 김지후(강수리영) right now, so 김지후(강수리영)'s hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 김지후(강수리영): its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 편의점 카운터에 서서 두툼한 돈봉투를 쥔 채 멍한 눈으로 허공을 응시하는 김지후(강수리영)의 모습.\n\nLOCATION (lock): Behind the convenience-store checkout counter on the sales floor at night. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Observe 김지후(강수리영) frontally from the prescribed eye-level position across the counter, using a slightly wide medium frame that includes her upper body and both hands around the thick money envelope. She sits near frame center behind the counter with vacant eyes fixed beyond the lens, while the envelope remains clearly readable in the lower middle; the emphasized axis is her static, isolated position.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: convenience-store counter (김지후 positioned behind it) — The customer-facing side and top edge are visible across the lower frame; used as Creates the horizontal foreground boundary separating the observer from 김지후; thick money envelope (held and being handled) — Its broad side is angled partly toward the camera between her hands; used as Held within the lower central portion of the composition as the physical focus of her restless hands.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Ambient convenience-store illumination renders the present space clearly while the restrained, high-contrast palette keeps her expression emotionally remote.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her white animal-bone necklace and holds the thick cash envelope at the counter.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 김지후(강수리영) right now, so 김지후(강수리영)'s hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 김지후(강수리영): its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 편의점 카운터에 서서 두툼한 돈봉투를 쥔 채 멍한 눈으로 허공을 응시하는 김지후(강수리영)의 모습.\n\nLOCATION (lock): Behind the convenience-store checkout counter on the sales floor at night. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Observe 김지후(강수리영) frontally from the prescribed eye-level position across the counter, using a slightly wide medium frame that includes her upper body and both hands around the thick money envelope. She sits near frame center behind the counter with vacant eyes fixed beyond the lens, while the envelope remains clearly readable in the lower middle; the emphasized axis is her static, isolated position.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: convenience-store counter (김지후 positioned behind it) — The customer-facing side and top edge are visible across the lower frame; used as Creates the horizontal foreground boundary separating the observer from 김지후; thick money envelope (held and being handled) — Its broad side is angled partly toward the camera between her hands; used as Held within the lower central portion of the composition as the physical focus of her restless hands.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Ambient convenience-store illumination renders the present space clearly while the restrained, high-contrast palette keeps her expression emotionally remote.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her white animal-bone necklace and holds the thick cash envelope at the counter.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 김지후(강수리영) right now, so 김지후(강수리영)'s hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 김지후(강수리영): its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S10sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S10sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:904317>"
    },
    {
     "label": "PROP REFERENCE — 두툼한 돈봉투: the exact object appearing in this shot; match its look, material and wear exactly.",
     "path": "<bytes:1264547>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L09B02.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:904317>"
    },
    {
     "label": "PROP REFERENCE — 두툼한 돈봉투: the exact object appearing in this shot; match its look, material and wear exactly.",
     "path": "<bytes:1264547>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 카운터 뒤에서 정면(카메라 축)을 향하고 있고, 시선은 렌즈를 관통해 허공으로 초점 없이 향한다. 얼굴·몸통·봉투 모두 카메라 정면. 봉투의 넓은 면이 카메라 쪽으로 각져 있어 지시대로다.",
      "built_space": "편의점 판매장 내부: 좌측 음료 냉장고, 중앙 진열대, 배경의 유리 자동문(야간 어두운 외부), 천장 형광등, CCTV 돔이 레퍼런스와 일치. 우측에는 POS 모니터·키보드·카드단말기가 놓인 계산대가 있고, 하단 전경으로 카운터 상판 모서리가 가로로 지나가 관찰자와 인물을 분리한다. 계산대는 하나뿐이며 인물은 카운터 뒤(점원 위치)에 서 있다.",
      "entities": "인물: 20대 초반 한국 여성, 긴 검은 생머리, 캐릭터 레퍼런스와 얼굴·헤어 유사. 회색 티셔츠(레퍼런스 이너와 동일 계열, 조끼는 미착용). 목에 흰 뼈 모양 펜던트 목걸이 착용 — 'white animal-bone necklace' 충족. 소품: 갈색 크라프트지 두툼한 돈봉투, 상단 접힘·테이프 흔적·'10000' 지폐 다발 자국까지 프롭 레퍼런스와 매우 근접. 화면 내 판독 가능한 캡션/워터마크 없음.",
      "hard_violations": [],
      "physics": "봉투는 카운터 상판 위에 세워져 놓이고 양손이 좌우에서 감싸 쥐고 있다 — 하단은 상판이 받치고 손이 측면을 지지. 팔뚝은 상판에 자연스럽게 얹혀 무게가 실린다. 떠 있는 물체나 지지 없는 신체 없음."
     },
     {
      "label": "B",
      "direction": "인물은 정면을 향하고 시선은 렌즈 약간 위/너머 허공을 응시. 봉투 넓은 면이 카메라 쪽으로 향함. 방향성은 요구대로.",
      "built_space": "동일 편의점 공간이지만 카운터 상판이 인물의 가슴 아래 매우 낮게, 그리고 화면 우하단에서 대각선으로 급격히 꺾여 인물 앞의 전경 경계가 약하게 읽힌다. 우측 계산대(모니터·키보드·단말기)는 별도로 존재해 계산대 중복은 아니나, 인물은 자기 앞 카운터와 우측 계산대 사이 애매한 위치. 우하단에 가격표/POP 판넬이 크게 전경으로 들어와 있고 그 위 문자·숫자('2500', '2,990' 등)가 판독 가능하게 삽입되어 있다. 배경 진열대 구조는 레퍼런스와 대체로 일치.",
      "entities": "인물: 20대 한국 여성, 긴 검은 머리, 회색 티셔츠 — 캐릭터 레퍼런스와 대체로 일치하나 얼굴 나이감이 다소 위로 보임. 목걸이는 흰 구슬/뼛조각을 여러 개 꿴 형태로 레퍼런스 계열에서 벗어나진 않음. 소품 봉투는 크라프트지이나 표면 인쇄·'10000' 자국·테이프 디테일이 약해 프롭 레퍼런스 재현도가 낮다. 화면에 다수의 한글/숫자 텍스트가 뚜렷이 보인다.",
      "hard_violations": [
       "프레임 우측 및 우하단에 판독 가능한 가격표·POP 문자·숫자가 크게 삽입되어 'No text, captions, annotations' 지시 위반 수준으로 노출"
      ],
      "physics": "봉투는 양손으로 좌우에서 감싸 쥐고 카운터 앞쪽에 세워져 있으나 하단이 상판 모서리에 걸친 듯 애매하게 접해 지지면이 불분명하다. 다만 두 손이 확실히 잡고 있어 부유 상태는 아니다. 인물의 몸도 지면에 정상 지지."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "카운터 뒤 정면 아이레벨 미디엄샷, 하단 전경의 카운터 경계, 양손이 감싼 두툼한 돈봉투(프롭 레퍼런스와 거의 동일), 흰 뼈 목걸이, 야간 편의점 로케이션 모두 충족하며 삽입 텍스트도 없다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "구도와 시선·소품 파지는 무난하나 우하단 가격 POP의 판독 가능한 문자·숫자 노출이 '텍스트 금지' 조항을 정면으로 어기고, 봉투의 재질·인쇄 디테일과 카운터 전경 경계 재현도가 A보다 떨어진다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "카운터 뒤 정면 아이레벨 미디엄샷, 하단 전경의 카운터 경계, 양손이 감싼 두툼한 돈봉투(프롭 레퍼런스와 거의 동일), 흰 뼈 목걸이, 야간 편의점 로케이션 모두 충족하며 삽입 텍스트도 없다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "구도와 시선·소품 파지는 무난하나 우하단 가격 POP의 판독 가능한 문자·숫자 노출이 '텍스트 금지' 조항을 정면으로 어기고, 봉투의 재질·인쇄 디테일과 카운터 전경 경계 재현도가 A보다 떨어진다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 카운터 뒤에서 정면(카메라 축)을 향하고 있고, 시선은 렌즈를 관통해 허공으로 초점 없이 향한다. 얼굴·몸통·봉투 모두 카메라 정면. 봉투의 넓은 면이 카메라 쪽으로 각져 있어 지시대로다.",
      "built_space": "편의점 판매장 내부: 좌측 음료 냉장고, 중앙 진열대, 배경의 유리 자동문(야간 어두운 외부), 천장 형광등, CCTV 돔이 레퍼런스와 일치. 우측에는 POS 모니터·키보드·카드단말기가 놓인 계산대가 있고, 하단 전경으로 카운터 상판 모서리가 가로로 지나가 관찰자와 인물을 분리한다. 계산대는 하나뿐이며 인물은 카운터 뒤(점원 위치)에 서 있다.",
      "entities": "인물: 20대 초반 한국 여성, 긴 검은 생머리, 캐릭터 레퍼런스와 얼굴·헤어 유사. 회색 티셔츠(레퍼런스 이너와 동일 계열, 조끼는 미착용). 목에 흰 뼈 모양 펜던트 목걸이 착용 — 'white animal-bone necklace' 충족. 소품: 갈색 크라프트지 두툼한 돈봉투, 상단 접힘·테이프 흔적·'10000' 지폐 다발 자국까지 프롭 레퍼런스와 매우 근접. 화면 내 판독 가능한 캡션/워터마크 없음.",
      "hard_violations": [],
      "physics": "봉투는 카운터 상판 위에 세워져 놓이고 양손이 좌우에서 감싸 쥐고 있다 — 하단은 상판이 받치고 손이 측면을 지지. 팔뚝은 상판에 자연스럽게 얹혀 무게가 실린다. 떠 있는 물체나 지지 없는 신체 없음."
     },
     {
      "label": "B",
      "direction": "인물은 정면을 향하고 시선은 렌즈 약간 위/너머 허공을 응시. 봉투 넓은 면이 카메라 쪽으로 향함. 방향성은 요구대로.",
      "built_space": "동일 편의점 공간이지만 카운터 상판이 인물의 가슴 아래 매우 낮게, 그리고 화면 우하단에서 대각선으로 급격히 꺾여 인물 앞의 전경 경계가 약하게 읽힌다. 우측 계산대(모니터·키보드·단말기)는 별도로 존재해 계산대 중복은 아니나, 인물은 자기 앞 카운터와 우측 계산대 사이 애매한 위치. 우하단에 가격표/POP 판넬이 크게 전경으로 들어와 있고 그 위 문자·숫자('2500', '2,990' 등)가 판독 가능하게 삽입되어 있다. 배경 진열대 구조는 레퍼런스와 대체로 일치.",
      "entities": "인물: 20대 한국 여성, 긴 검은 머리, 회색 티셔츠 — 캐릭터 레퍼런스와 대체로 일치하나 얼굴 나이감이 다소 위로 보임. 목걸이는 흰 구슬/뼛조각을 여러 개 꿴 형태로 레퍼런스 계열에서 벗어나진 않음. 소품 봉투는 크라프트지이나 표면 인쇄·'10000' 자국·테이프 디테일이 약해 프롭 레퍼런스 재현도가 낮다. 화면에 다수의 한글/숫자 텍스트가 뚜렷이 보인다.",
      "hard_violations": [
       "프레임 우측 및 우하단에 판독 가능한 가격표·POP 문자·숫자가 크게 삽입되어 'No text, captions, annotations' 지시 위반 수준으로 노출"
      ],
      "physics": "봉투는 양손으로 좌우에서 감싸 쥐고 카운터 앞쪽에 세워져 있으나 하단이 상판 모서리에 걸친 듯 애매하게 접해 지지면이 불분명하다. 다만 두 손이 확실히 잡고 있어 부유 상태는 아니다. 인물의 몸도 지면에 정상 지지."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "카운터 너머 아이레벨 정면 미디엄 프레임, 하단 중앙의 두툼한 돈봉투를 양손이 감싼 상태, 흰 뼈 목걸이와 야간 편의점 로케이션 일치까지 모두 충족하며 멍한 시선 연기가 가장 설득력 있음."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "프레이밍과 소품 재현은 준수하나 목걸이가 레퍼런스의 흰 동물뼈 목걸이가 아닌 평면적 만화풍 뼈 펜던트로 바뀌었고 카운터 상판 접합과 우측 구조 묘사가 흐트러져 A보다 뒤짐."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.778,
      "B": 1.75
     },
     "adjusted": {
      "A": 1.778,
      "B": 1.75
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "B"
     },
     "agreed": false
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "카운터 너머 아이레벨 정면 미디엄 프레임, 하단 중앙의 두툼한 돈봉투를 양손이 감싼 상태, 흰 뼈 목걸이와 야간 편의점 로케이션 일치까지 모두 충족하며 멍한 시선 연기가 가장 설득력 있음."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍과 소품 재현은 준수하나 목걸이가 레퍼런스의 흰 동물뼈 목걸이가 아닌 평면적 만화풍 뼈 펜던트로 바뀌었고 카운터 상판 접합과 우측 구조 묘사가 흐트러져 A보다 뒤짐."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 13
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "readings": [
   {
    "label": "A",
    "direction": "김지후는 카운터 뒤에서 정면(카메라 축)을 향하고 있고, 시선은 렌즈를 관통해 허공으로 초점 없이 향한다. 얼굴·몸통·봉투 모두 카메라 정면. 봉투의 넓은 면이 카메라 쪽으로 각져 있어 지시대로다.",
    "built_space": "편의점 판매장 내부: 좌측 음료 냉장고, 중앙 진열대, 배경의 유리 자동문(야간 어두운 외부), 천장 형광등, CCTV 돔이 레퍼런스와 일치. 우측에는 POS 모니터·키보드·카드단말기가 놓인 계산대가 있고, 하단 전경으로 카운터 상판 모서리가 가로로 지나가 관찰자와 인물을 분리한다. 계산대는 하나뿐이며 인물은 카운터 뒤(점원 위치)에 서 있다.",
    "entities": "인물: 20대 초반 한국 여성, 긴 검은 생머리, 캐릭터 레퍼런스와 얼굴·헤어 유사. 회색 티셔츠(레퍼런스 이너와 동일 계열, 조끼는 미착용). 목에 흰 뼈 모양 펜던트 목걸이 착용 — 'white animal-bone necklace' 충족. 소품: 갈색 크라프트지 두툼한 돈봉투, 상단 접힘·테이프 흔적·'10000' 지폐 다발 자국까지 프롭 레퍼런스와 매우 근접. 화면 내 판독 가능한 캡션/워터마크 없음.",
    "hard_violations": [],
    "physics": "봉투는 카운터 상판 위에 세워져 놓이고 양손이 좌우에서 감싸 쥐고 있다 — 하단은 상판이 받치고 손이 측면을 지지. 팔뚝은 상판에 자연스럽게 얹혀 무게가 실린다. 떠 있는 물체나 지지 없는 신체 없음."
   },
   {
    "label": "B",
    "direction": "인물은 정면을 향하고 시선은 렌즈 약간 위/너머 허공을 응시. 봉투 넓은 면이 카메라 쪽으로 향함. 방향성은 요구대로.",
    "built_space": "동일 편의점 공간이지만 카운터 상판이 인물의 가슴 아래 매우 낮게, 그리고 화면 우하단에서 대각선으로 급격히 꺾여 인물 앞의 전경 경계가 약하게 읽힌다. 우측 계산대(모니터·키보드·단말기)는 별도로 존재해 계산대 중복은 아니나, 인물은 자기 앞 카운터와 우측 계산대 사이 애매한 위치. 우하단에 가격표/POP 판넬이 크게 전경으로 들어와 있고 그 위 문자·숫자('2500', '2,990' 등)가 판독 가능하게 삽입되어 있다. 배경 진열대 구조는 레퍼런스와 대체로 일치.",
    "entities": "인물: 20대 한국 여성, 긴 검은 머리, 회색 티셔츠 — 캐릭터 레퍼런스와 대체로 일치하나 얼굴 나이감이 다소 위로 보임. 목걸이는 흰 구슬/뼛조각을 여러 개 꿴 형태로 레퍼런스 계열에서 벗어나진 않음. 소품 봉투는 크라프트지이나 표면 인쇄·'10000' 자국·테이프 디테일이 약해 프롭 레퍼런스 재현도가 낮다. 화면에 다수의 한글/숫자 텍스트가 뚜렷이 보인다.",
    "hard_violations": [
     "프레임 우측 및 우하단에 판독 가능한 가격표·POP 문자·숫자가 크게 삽입되어 'No text, captions, annotations' 지시 위반 수준으로 노출"
    ],
    "physics": "봉투는 양손으로 좌우에서 감싸 쥐고 카운터 앞쪽에 세워져 있으나 하단이 상판 모서리에 걸친 듯 애매하게 접해 지지면이 불분명하다. 다만 두 손이 확실히 잡고 있어 부유 상태는 아니다. 인물의 몸도 지면에 정상 지지."
   }
  ],
  "totals": {
   "A": 14,
   "B": 13
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "카운터 뒤 정면 아이레벨 미디엄샷, 하단 전경의 카운터 경계, 양손이 감싼 두툼한 돈봉투(프롭 레퍼런스와 거의 동일), 흰 뼈 목걸이, 야간 편의점 로케이션 모두 충족하며 삽입 텍스트도 없다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "구도와 시선·소품 파지는 무난하나 우하단 가격 POP의 판독 가능한 문자·숫자 노출이 '텍스트 금지' 조항을 정면으로 어기고, 봉투의 재질·인쇄 디테일과 카운터 전경 경계 재현도가 A보다 떨어진다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L09B02.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:904317>"
   },
   {
    "label": "PROP REFERENCE — 두툼한 돈봉투: the exact object appearing in this shot; match its look, material and wear exactly.",
    "path": "<bytes:1264547>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "[Critical] 인물이 렌즈 너머 허공을 멍하게 응시해야 한다는 지시와 렌즈 정면 응시 금지 규칙을 어기고 카메라 렌즈를 똑바로 바라보고 있어 영화적 숏의 의도를 훼손합니다.",
     "fix_en": "Redraw only the character's eyes to shift her pupils slightly off-center, directing her vacant gaze into the void beyond the camera lens. Preserve exactly the person present and her position, her t-shirt and bone necklace, the money envelope, her hands, the counter, the background set, the lighting, and the framing."
    },
    {
     "issue_ko": "[Major] 돈봉투의 넓은 면이 카메라를 향해 약간 비스듬히 틀어져 있어야 한다는 지시와 달리, 렌즈를 향해 완전히 평행하게 정면으로 놓여 있습니다.",
     "fix_en": ""
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 5,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw only the character's eyes to shift her pupils slightly off-center, directing her vacant gaze into the void beyond the camera lens. Preserve exactly the person present and her position, her t-shirt and bone necklace, the money envelope, her hands, the counter, the background set, the lighting, and the framing.\n- \n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "프레이밍·시선·봉투 배치·목걸이 모두 계약대로이며, 좌측 매대 상품 진열이 배경 레퍼런스의 밀도를 더 잘 유지해 LOCATION 락에 근접하다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "인물·프롭·구도는 A와 동일하나, 계약이 요구하지 않은 좌측 매대·냉장고 진열 상품의 대량 소실(빈 선반 노출)로 배경 고정 조건이 후퇴한 손실만 남았다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.75
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.75
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "프레이밍·시선·봉투 배치·목걸이 모두 계약대로이며, 좌측 매대 상품 진열이 배경 레퍼런스의 밀도를 더 잘 유지해 LOCATION 락에 근접하다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "인물·프롭·구도는 A와 동일하나, 계약이 요구하지 않은 좌측 매대·냉장고 진열 상품의 대량 소실(빈 선반 노출)로 배경 고정 조건이 후퇴한 손실만 남았다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "정면 아이레벨 미디엄 숏, 카운터 뒤 중앙 배치, 두 손으로 쥔 크라프트 돈봉투, 흰 뼈 목걸이까지 계약 항목을 모두 충족하고 배경 레퍼런스의 진열 상품·집기 디테일도 가장 근접하게 보존."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "구도·시선·프롭은 A와 사실상 동일하나, 계약이 요구하지 않은 좌측 진열대 상품 재배치와 봉투 테이프·목걸이 체인 디테일 약화가 발생해 배경 락과 프롭 일치도에서 손해만 남음."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.653
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.653
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "정면 아이레벨 미디엄 숏, 카운터 뒤 중앙 배치, 두 손으로 쥔 크라프트 돈봉투, 흰 뼈 목걸이까지 계약 항목을 모두 충족하고 배경 레퍼런스의 진열 상품·집기 디테일도 가장 근접하게 보존."
     },
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "구도·시선·프롭은 A와 사실상 동일하나, 계약이 요구하지 않은 좌측 진열대 상품 재배치와 봉투 테이프·목걸이 체인 디테일 약화가 발생해 배경 락과 프롭 일치도에서 손해만 남음."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 15
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S10sh1__bgfirst_bg.png",
   "bg_asset_id": "aded442b-127e-4feb-8b3f-28bbaf0e4913",
   "bg_record_key": "S10sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S11sh4": {
  "input_fingerprint": "75594e91d666f28d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 다세대 주택의 낡은 외부 계단을 다급히 오르며 한쪽 발로 위칸을 딛고 반대쪽 발이 허공에 떠 있는 김지후(강수리영)와 정인우의 mid-action 뒷모습.\n\nLOCATION (lock): The worn outdoor staircase attached to the multi-family house, rising from the dark alley. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track upward from lower-torso height two to three meters behind 김지후(강수리영) and 정인우, angling along the external staircase as both climb away from the camera. Freeze the keyframe with each figure transferring weight onto a higher step while the opposite foot hangs free, placing them on offset diagonals so neither body overlaps the other; the emphasized change is their upward screen movement.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, midground, moves toward upper external stairs; 정인우 in the middle-right of the frame, midground, moves toward upper external stairs; upper external stairs in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: external staircase (old) — The stair run rises from the lower foreground toward the upper background; used as Creates the rising diagonal that carries both figures away from the camera; stair railing (fixed alongside the staircase) — Its length follows the upward angle of the stairs; used as Provides the side boundary around the two ascending figures.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Dark, low-key nighttime illumination silhouettes the urgent leg movement while retaining enough separation to read both bodies.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same narrow residential alley, worn exterior stairs, darkness, and red emergency-light spill. Exclude the stopped-car focal action; show the two figures urgently climbing the stairs.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu continues to wear the white animal-bone necklace and carries her phone while climbing with Inwoo.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face); 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 3,
   "C": 8
  },
  "selected": "C",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "카메라 각도·오프셋 대각선·한 발 접지/한 발 공중의 mid-action이 정확하고 지후 의상(파란 조끼)도 락에 맞으나, 인우의 회색 비니 부재와 이전 스틸의 붉은 비상등 스필이 거의 사라진 점이 감점."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "제외 지시된 정차 경찰차를 프레임 중앙에 남겼고, 지후가 단발·검은 상의로 캐릭터 레퍼런스(긴 생머리·파란 조끼)와 어긋나며 철제 계단이 로케이션 락과도 불일치.  ★위반: [claude-opus] PREVIOUS STILL USAGE가 명시적으로 제외하라고 한 정차 경찰차(정차 차량 초점 액션)를 프레임 중앙 하단에 그대로 넣었다 — 지시된 제외 요소의 재도입"
   },
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "낡은 외부 계단 상승 대각선, 뒤 2~3m 하부 몸통 높이 뒷모습, 두 인물의 한 발 접지·한 발 공중, 좁은 골목의 붉은 비상등 스필, 뼈 목걸이·휴대폰 소지까지 모두 충족 — 지후 상의만 이전 스틸 조끼와 다른 점이 유일한 감점."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S11sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:904317>"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:871658>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "김지후가 기준 이미지의 파란색 조끼와 회색 반팔 티셔츠 대신 검은색 긴팔 상의를 입고 있습니다.",
     "fix_en": "Change Jihu's upper body clothing from the black long-sleeved top to a blue safety vest worn open over a grey short-sleeved t-shirt, showing her bare lower arms. Preserve her exact pose, hair, pants, the phone in her hand, Inwoo's appearance and pose, the staircase, the lighting, and the framing.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "동물의 뼈 목걸이가 가슴 앞쪽이 아닌 등 뒤에 비정상적으로 매달려 있습니다.",
     "fix_en": "Remove the white bone necklace from Jihu's back so that only the back of her clothing is visible there. Preserve her exact pose, hair, pants, Inwoo's appearance, the staircase, the lighting, and the framing exactly.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change Jihu's upper body clothing from the black long-sleeved top to a blue safety vest worn open over a grey short-sleeved t-shirt, showing her bare lower arms. Preserve her exact pose, hair, pants, the phone in her hand, Inwoo's appearance and pose, the staircase, the lighting, and the framing.\n- Remove the white bone necklace from Jihu's back so that only the back of her clothing is visible there. Preserve her exact pose, hair, pants, Inwoo's appearance, the staircase, the lighting, and the framing exactly.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·계단 대각선·발 들림 등 스테이징은 충족하나, 계약이 요구하지 않은 김지후의 의상 교체(레퍼런스의 파란 형광 조끼 → 전신 검정 후드)와 목 아닌 등 뒤에 걸린 기이한 뼈 목걸이 배치가 정체성·의상 잠금을 훼손한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 카메라 축·와이드 뒷모습·오프셋 대각선과 체중 이동을 유지하면서 김지후의 파란 작업 조끼 복장을 레퍼런스대로 복원했고 장소·조명·인물 수 모두 잠금과 일치 — 뼈 목걸이 가시성이 낮아진 점만 손실이다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.625,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.625,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·계단 대각선·발 들림 등 스테이징은 충족하나, 계약이 요구하지 않은 김지후의 의상 교체(레퍼런스의 파란 형광 조끼 → 전신 검정 후드)와 목 아닌 등 뒤에 걸린 기이한 뼈 목걸이 배치가 정체성·의상 잠금을 훼손한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 카메라 축·와이드 뒷모습·오프셋 대각선과 체중 이동을 유지하면서 김지후의 파란 작업 조끼 복장을 레퍼런스대로 복원했고 장소·조명·인물 수 모두 잠금과 일치 — 뼈 목걸이 가시성이 낮아진 점만 손실이다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 인물 모두 카메라에 등을 보인 채 화면 상단(계단 위쪽)을 향해 오른다. 지후는 중앙 좌측, 인우는 중앙 우측 오프셋 대각선에 배치되어 서로 겹치지 않는다. 지후의 시선은 오른손에 든 휴대폰 화면 쪽(자기 눈 방향)으로 향하고, 폰의 화면 면은 그녀 쪽을 향하며 카메라는 측면/뒷면 각도를 본다 — 프롭 방향 정상. 계단 대각선은 하단 전경에서 상단 배경으로 상승.",
      "built_space": "외부 철제 난간이 계단 양옆을 따라 상승, 상단 배경으로 층계참과 건물 창(따뜻한 실내광)이 이어진다. 좌측은 좁은 골목과 젖은 아스팔트, 붉은 경광등 스필이 벽에 번져 이전 스틸의 장소·재질과 일치. 계단 단 수와 난간 연속성 정상, 중복 구조물 없음.",
      "entities": "지후: 긴 검은 머리의 한국인 여성, 캐릭터 레퍼런스와 동일한 파란 반사 조끼+회색 티+짙은 카고 바지+워크부츠 — 의상 락 일치. 다만 흰 동물뼈 목걸이는 뒷모습이라 목뒤로 가려 보이지 않음. 인우: 회색 비니, 네이비 재킷, 회색 바지, 짙은 슈즈 — 레퍼런스 일치. 제3의 인물 없음, 텍스트/워터마크 없음.",
      "hard_violations": [],
      "physics": "지후는 왼발이 위칸을 디디고 뒤쪽 발은 아래 단에서 막 떨어지는 중, 왼손은 난간을 잡아 지지. 인우는 오른발이 위칸에 실려 있고 뒤쪽 발이 공중에 뜬 상태 — 체중 이동이 명확하며 지지점(디딘 발, 난간)이 모두 확인됨. 공중에 뜬 무지지 신체 없음."
     },
     {
      "label": "B",
      "direction": "동일하게 두 인물이 등을 보이고 계단 위로 오른다. 지후 중앙 좌측, 인우 중앙 우측. 지후는 폰을 오른손에 들고 화면이 자기 쪽을 향함 — 방향 정상. 상승 대각선 동일.",
      "built_space": "A와 사실상 동일한 계단·난간·골목·붉은 스필. 다만 우측 하단 난간 영역과 계단 텍스처가 다소 어둡게 뭉개져 배경 창의 디테일이 줄었고, 오른쪽 아래 프레임 가장자리에 검은 띠 형태의 크롭 흔적이 보인다.",
      "entities": "지후의 의상이 레퍼런스의 파란 형광 조끼가 아닌 전신 검정 상의로 바뀜 — 의상 락 위반 수준의 임의 변경. 대신 목 뒤로 흰 뼈 목걸이가 등 쪽에 늘어져 보이는데, 목걸이가 등 뒤에 매달린 형태는 착용 방식으로 부자연스럽고 스티커처럼 붙어 보인다. 인우는 A와 동일. 제3 인물 없음.",
      "hard_violations": [],
      "physics": "인우는 A와 동일하게 앞발이 위칸, 뒷발이 공중. 지후는 왼손이 난간에 닿아 있고 발은 위칸을 딛는 중이나, A보다 뒷발 들림이 덜 읽히고 다리 실루엣이 어둠에 묻혀 '한쪽 발 허공' 순간이 약하게 전달된다. 무지지 부유체는 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지후의 파란 안전조끼 등 레퍼런스 의상 락을 지키고, 두 인물 모두 위칸을 디딘 발과 허공에 뜬 반대발이 명확히 읽히며 오프셋 대각선·와이드·뒷모습 프레이밍을 정확히 충족한다. 뼈 목걸이가 뒷모습에 가려 보이지 않는 점만 아쉽다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "뼈 목걸이를 보이게 하려다 지후의 잠긴 의상(파란 형광 조끼+회색 티)을 통째로 검정 상의로 바꿔버려 계약이 요구하지 않은 정체성·의상 손실이 크고, 목걸이가 등 뒤에 매달린 부자연스러운 착용에 다리 동작도 어둠에 묻혀 mid-action이 약해졌다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지후의 파란 안전조끼 등 레퍼런스 의상 락을 지키고, 두 인물 모두 위칸을 디딘 발과 허공에 뜬 반대발이 명확히 읽히며 오프셋 대각선·와이드·뒷모습 프레이밍을 정확히 충족한다. 뼈 목걸이가 뒷모습에 가려 보이지 않는 점만 아쉽다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "뼈 목걸이를 보이게 하려다 지후의 잠긴 의상(파란 형광 조끼+회색 티)을 통째로 검정 상의로 바꿔버려 계약이 요구하지 않은 정체성·의상 손실이 크고, 목걸이가 등 뒤에 매달린 부자연스러운 착용에 다리 동작도 어둠에 묻혀 mid-action이 약해졌다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "두 인물 모두 카메라에 등을 보인 채 화면 상단(계단 위쪽)을 향해 오른다. 지후는 중앙 좌측, 인우는 중앙 우측 오프셋 대각선에 배치되어 서로 겹치지 않는다. 지후의 시선은 오른손에 든 휴대폰 화면 쪽(자기 눈 방향)으로 향하고, 폰의 화면 면은 그녀 쪽을 향하며 카메라는 측면/뒷면 각도를 본다 — 프롭 방향 정상. 계단 대각선은 하단 전경에서 상단 배경으로 상승.",
      "built_space": "외부 철제 난간이 계단 양옆을 따라 상승, 상단 배경으로 층계참과 건물 창(따뜻한 실내광)이 이어진다. 좌측은 좁은 골목과 젖은 아스팔트, 붉은 경광등 스필이 벽에 번져 이전 스틸의 장소·재질과 일치. 계단 단 수와 난간 연속성 정상, 중복 구조물 없음.",
      "entities": "지후: 긴 검은 머리의 한국인 여성, 캐릭터 레퍼런스와 동일한 파란 반사 조끼+회색 티+짙은 카고 바지+워크부츠 — 의상 락 일치. 다만 흰 동물뼈 목걸이는 뒷모습이라 목뒤로 가려 보이지 않음. 인우: 회색 비니, 네이비 재킷, 회색 바지, 짙은 슈즈 — 레퍼런스 일치. 제3의 인물 없음, 텍스트/워터마크 없음.",
      "hard_violations": [],
      "physics": "지후는 왼발이 위칸을 디디고 뒤쪽 발은 아래 단에서 막 떨어지는 중, 왼손은 난간을 잡아 지지. 인우는 오른발이 위칸에 실려 있고 뒤쪽 발이 공중에 뜬 상태 — 체중 이동이 명확하며 지지점(디딘 발, 난간)이 모두 확인됨. 공중에 뜬 무지지 신체 없음."
     },
     {
      "label": "A",
      "direction": "동일하게 두 인물이 등을 보이고 계단 위로 오른다. 지후 중앙 좌측, 인우 중앙 우측. 지후는 폰을 오른손에 들고 화면이 자기 쪽을 향함 — 방향 정상. 상승 대각선 동일.",
      "built_space": "A와 사실상 동일한 계단·난간·골목·붉은 스필. 다만 우측 하단 난간 영역과 계단 텍스처가 다소 어둡게 뭉개져 배경 창의 디테일이 줄었고, 오른쪽 아래 프레임 가장자리에 검은 띠 형태의 크롭 흔적이 보인다.",
      "entities": "지후의 의상이 레퍼런스의 파란 형광 조끼가 아닌 전신 검정 상의로 바뀜 — 의상 락 위반 수준의 임의 변경. 대신 목 뒤로 흰 뼈 목걸이가 등 쪽에 늘어져 보이는데, 목걸이가 등 뒤에 매달린 형태는 착용 방식으로 부자연스럽고 스티커처럼 붙어 보인다. 인우는 A와 동일. 제3 인물 없음.",
      "hard_violations": [],
      "physics": "인우는 A와 동일하게 앞발이 위칸, 뒷발이 공중. 지후는 왼손이 난간에 닿아 있고 발은 위칸을 딛는 중이나, A보다 뒷발 들림이 덜 읽히고 다리 실루엣이 어둠에 묻혀 '한쪽 발 허공' 순간이 약하게 전달된다. 무지지 부유체는 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S11sh2"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S12sh6": {
  "input_fingerprint": "560d697f4dc7dc41",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 고산들의 앞을 막아선 채 손에 쥔 삼단봉을 허공에 높이 치켜든 정인우의 찰나.\n\nLOCATION (lock): The central floor of the wrecked living room inside the multi-family home. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From chest height on 정인우's three-quarter side, tilt upward with his extended baton as he blocks 고산들 at medium distance. 정인우 occupies the left foreground with his raised arm forming a diagonal above them, while 고산들 remains directly beyond him in the center-right midground, his attention cutting past the barrier toward 김지후 off-screen; the emphasized change is 정인우's raised defensive posture.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 정인우 in the middle-left of the frame, foreground; 고산들 in the middle-right of the frame, midground.\n- KEY BACKGROUND ELEMENTS: extended baton (extended and raised) — The baton rises diagonally from 정인우's hand into the upper frame; used as Its upward diagonal motivates the camera tilt and marks 정인우's temporary defensive line; broken furniture (broken and disordered) — Damaged sides and edges appear at varied angles behind the two men; used as Broken shapes remain visible behind the confrontation as spatial evidence of the room's condition; bloodstains (clearly visible); used as Visible behind and beneath the confrontation without obscuring the figures.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient illumination appropriate to the nighttime interior gives the raised baton and opposing profiles hard tonal separation against the disordered room.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same wrecked living room, broken furnishings, bloodstains, and lighting. Exclude any passive standing tableau; show the man blocking the intruder with an extended baton.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The wrecked, bloodstained living room remains unchanged around them. Jihu still wears her white animal-bone necklace.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 정인우 right now, so 정인우's hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 정인우: its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "C": 6,
   "B": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "삼단봉 단일·상향 대각선, 정인우 좌측 전경/고산들 우측 중경, 시선이 장벽 너머 오프스크린을 향하는 지시까지 충족하며 이전 스틸의 도어락·식탁·액자·혈흔 등 로케이션 락이 가장 정확하다."
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "지지와 인물 배치는 정상이나 봉이 '높이 치켜든' 대각선이 아니라 거의 수평 겨눔이고 고산들의 시선이 오프스크린이 아닌 정인우에게 고정, 로케이션 고정 요소(도어락·책장)가 다수 누락됐다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "인물 유사도는 가장 좋으나 삼단봉이 두 자루로 중복 렌더링되고 파지 물리가 성립하지 않아 하드 위반이며, 고산들의 시선도 오프스크린이 아닌 정면 대치로 어긋난다.  ★위반: [claude-opus] 삼단봉이 두 개로 중복 등장(발명된 소품/중복 제어물) / [claude-opus] 머리 위 봉과 전방 봉의 축이 손·팔과 정합되지 않아 그립 해부가 불가능하게 읽힘 / [gpt] 단 하나여야 하는 삼단봉이 두 개로 복제되어 정인우가 양손에 각각 들고 있다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S12sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:871658>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "정인우의 왼손 해부가 완전히 틀렸으며(오른손이 뒤집혀 붙은 형태), 삼단봉이 앞을 막지 않고 등 뒤(왼쪽 위)를 향해 거꾸로 뻗어 있습니다.",
     "fix_en": "Redraw Jung In-woo's raised left hand and the baton so that a correctly anatomized left hand firmly grips the handle, with the long extended shaft of the baton pointing forward and diagonally upward to the right to block the other man. Preserve Jung In-woo's head, body, and navy clothing, the man standing on the right, the wrecked living room set, the bloodstains, and the lighting exactly.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw Jung In-woo's raised left hand and the baton so that a correctly anatomized left hand firmly grips the handle, with the long extended shaft of the baton pointing forward and diagonally upward to the right to block the other man. Preserve Jung In-woo's head, body, and navy clothing, the man standing on the right, the wrecked living room set, the bloodstains, and the lighting exactly.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "정인우는 왼쪽 전경에서 3/4 측면으로 서 있고, 오른팔을 위로 들어 삼단봉이 프레임 좌상단으로 대각선 상승한다. 봉의 방향은 위·왼쪽 허공(천장 방향)으로, '허공에 높이 치켜든' 텍스트와 일치한다. 고산들은 중우측 중경에 서서 시선이 정인우를 지나 카메라 좌측/오프스크린 쪽을 향해 째려보는 각도로, 장벽 너머를 응시하는 지시와 부합한다.",
      "built_space": "전경 좌측 냉장고와 도어록, 좌측 벽 액자·메모지, 천장 트랙조명, 발코니 창, 에어컨, 소파, 넘어진 액자(핏자국), 우측 책장·바구니, 쓰러진 화분과 흙 — 이전 스틸의 고정물 배치가 거의 그대로 유지된다. 소품 중복이나 추가 인물 없음.",
      "entities": "정인우: 20대 중반 한국 남성, 검은 단발머리, 네이비 재킷+녹색 니트 소매가 보여 캐릭터 레퍼런스와 일치(비니는 없으나 이전 스틸에도 남자 미등장). 고산들: 30~40대 한국 남성, 어두운 재킷, 침입자 인상 — 샷 텍스트가 요구한 인물. 삼단봉: 금속 신축봉으로 손잡이 그립이 손에 잡혀 있고 마디가 보인다.",
      "hard_violations": [],
      "physics": "정인우는 두 발이 바닥에 있고 상체가 약간 앞으로 기울며 팔을 위로 뻗어 봉의 무게를 지탱한다. 봉은 주먹에 확실히 쥐어져 있고 그립 부분이 손 안에 있다. 고산들은 한 발을 앞으로 내딛는 중량 이동이 보이며 바닥 접지. 공중에 뜬 무지지 물체 없음."
     },
     {
      "label": "B",
      "direction": "정인우의 손은 A와 같은 위치에서 봉의 그립을 쥐고 있으나, 봉이 손에서 오른쪽 아래로 길게 뻗어 고산들 쪽 바닥/무릎 높이까지 내려가고, 동시에 또 하나의 긴 금속 막대가 우상단에서 좌하단으로 교차하듯 화면을 가로지른다. 즉 봉이 '허공에 높이 치켜든' 상승 대각선이 아니라 하강 대각선이며, 두 개의 막대가 X자로 겹쳐 보인다. 고산들의 시선은 A와 동일.",
      "built_space": "배경 고정물(냉장고, 도어록, 트랙조명, 창, 소파, 넘어진 액자, 책장, 화분)은 A/이전 스틸과 동일. 그러나 프레임을 관통하는 금속 봉이 실질적으로 2개(손에서 내려가는 것 + 우상단에서 오는 것)로 읽혀 도구 개수가 잘못되었다.",
      "entities": "정인우·고산들의 얼굴·의상은 A와 거의 동일. 삼단봉은 손잡이 마디가 보이지만 봉신이 비정상적으로 길고, 별도의 두 번째 막대가 존재해 '삼단봉 하나'라는 소품 규정을 위반.",
      "hard_violations": [
       "손에 쥐어지지 않은 두 번째 긴 금속 봉이 프레임을 가로질러 떠 있음 — 지지물·손 없음(소품 중복/부유 물체)",
       "샷 텍스트의 '허공에 높이 치켜든' 상승 대각선과 반대로 봉이 아래쪽으로 향함(방향 위반)"
      ],
      "physics": "정인우의 발과 자세는 A와 동일하게 접지. 그러나 우상단에서 화면을 가로지르는 막대는 어느 손도 잡고 있지 않고 벽·가구에도 걸쳐 있지 않아 아무것도 지탱하지 않는다. 손에 쥔 봉 역시 무게중심상 팔을 들어올린 자세와 어긋나 아래로 처져 있어 '치켜든' 동작이 성립하지 않는다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "삼단봉이 손에 쥐어진 채 좌상단으로 치켜올려져 '허공에 높이 치켜든' 순간과 카메라 틸트 동기를 정확히 구현했고, 좌전경 정인우/우중경 고산들 배치·야간 저조도·파손된 거실 고정물까지 이전 스틸과 일치한다."
     },
     {
      "label": "B",
      "score": 2,
      "verdict_ko": "봉이 위가 아니라 아래로 향해 '치켜든' 지시를 뒤집었고, 아무도 잡지 않은 두 번째 금속 막대가 화면을 가로질러 떠 있어 소품 중복·부유 물체라는 치명적 결함을 추가했다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "삼단봉이 손에 쥐어진 채 좌상단으로 치켜올려져 '허공에 높이 치켜든' 순간과 카메라 틸트 동기를 정확히 구현했고, 좌전경 정인우/우중경 고산들 배치·야간 저조도·파손된 거실 고정물까지 이전 스틸과 일치한다."
     },
     {
      "label": "B",
      "score": 2,
      "verdict_ko": "봉이 위가 아니라 아래로 향해 '치켜든' 지시를 뒤집었고, 아무도 잡지 않은 두 번째 금속 막대가 화면을 가로질러 떠 있어 소품 중복·부유 물체라는 치명적 결함을 추가했다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "정인우는 왼쪽 전경에서 3/4 측면으로 서 있고, 오른팔을 위로 들어 삼단봉이 프레임 좌상단으로 대각선 상승한다. 봉의 방향은 위·왼쪽 허공(천장 방향)으로, '허공에 높이 치켜든' 텍스트와 일치한다. 고산들은 중우측 중경에 서서 시선이 정인우를 지나 카메라 좌측/오프스크린 쪽을 향해 째려보는 각도로, 장벽 너머를 응시하는 지시와 부합한다.",
      "built_space": "전경 좌측 냉장고와 도어록, 좌측 벽 액자·메모지, 천장 트랙조명, 발코니 창, 에어컨, 소파, 넘어진 액자(핏자국), 우측 책장·바구니, 쓰러진 화분과 흙 — 이전 스틸의 고정물 배치가 거의 그대로 유지된다. 소품 중복이나 추가 인물 없음.",
      "entities": "정인우: 20대 중반 한국 남성, 검은 단발머리, 네이비 재킷+녹색 니트 소매가 보여 캐릭터 레퍼런스와 일치(비니는 없으나 이전 스틸에도 남자 미등장). 고산들: 30~40대 한국 남성, 어두운 재킷, 침입자 인상 — 샷 텍스트가 요구한 인물. 삼단봉: 금속 신축봉으로 손잡이 그립이 손에 잡혀 있고 마디가 보인다.",
      "hard_violations": [],
      "physics": "정인우는 두 발이 바닥에 있고 상체가 약간 앞으로 기울며 팔을 위로 뻗어 봉의 무게를 지탱한다. 봉은 주먹에 확실히 쥐어져 있고 그립 부분이 손 안에 있다. 고산들은 한 발을 앞으로 내딛는 중량 이동이 보이며 바닥 접지. 공중에 뜬 무지지 물체 없음."
     },
     {
      "label": "B",
      "direction": "정인우의 손은 A와 같은 위치에서 봉의 그립을 쥐고 있으나, 봉이 손에서 오른쪽 아래로 길게 뻗어 고산들 쪽 바닥/무릎 높이까지 내려가고, 동시에 또 하나의 긴 금속 막대가 우상단에서 좌하단으로 교차하듯 화면을 가로지른다. 즉 봉이 '허공에 높이 치켜든' 상승 대각선이 아니라 하강 대각선이며, 두 개의 막대가 X자로 겹쳐 보인다. 고산들의 시선은 A와 동일.",
      "built_space": "배경 고정물(냉장고, 도어록, 트랙조명, 창, 소파, 넘어진 액자, 책장, 화분)은 A/이전 스틸과 동일. 그러나 프레임을 관통하는 금속 봉이 실질적으로 2개(손에서 내려가는 것 + 우상단에서 오는 것)로 읽혀 도구 개수가 잘못되었다.",
      "entities": "정인우·고산들의 얼굴·의상은 A와 거의 동일. 삼단봉은 손잡이 마디가 보이지만 봉신이 비정상적으로 길고, 별도의 두 번째 막대가 존재해 '삼단봉 하나'라는 소품 규정을 위반.",
      "hard_violations": [
       "손에 쥐어지지 않은 두 번째 긴 금속 봉이 프레임을 가로질러 떠 있음 — 지지물·손 없음(소품 중복/부유 물체)",
       "샷 텍스트의 '허공에 높이 치켜든' 상승 대각선과 반대로 봉이 아래쪽으로 향함(방향 위반)"
      ],
      "physics": "정인우의 발과 자세는 A와 동일하게 접지. 그러나 우상단에서 화면을 가로지르는 막대는 어느 손도 잡고 있지 않고 벽·가구에도 걸쳐 있지 않아 아무것도 지탱하지 않는다. 손에 쥔 봉 역시 무게중심상 팔을 들어올린 자세와 어긋나 아래로 처져 있어 '치켜든' 동작이 성립하지 않는다."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "정인우는 좌측 전경에서 우측(고산들 방향)을 향해 옆얼굴을 보이며 서 있고, 오른손은 위로 들려 있다. 그러나 손에서 뻗은 봉이 하나가 아니라 두 개의 긴 금속 막대로 보이며, 하나는 오른쪽 위 천장 쪽으로, 다른 하나는 오른쪽 아래(고산들 발치 쪽)로 X자를 그리며 프레임을 가로지른다. 고산들은 중우측 중경에 서서 화면 좌측(정인우/오프스크린 김지후 방향)으로 시선을 보내고 있어 시선 방향 자체는 맞다.",
      "built_space": "이전 스틸의 거실 구조(현관 쪽 벽·도어록, 좌측 테이블, 정면 발코니 창, 우측 소파와 넘어진 액자·리클라이너, 우측 책장, 바닥의 화분과 흙)가 대체로 유지된다. 다만 상단을 가로지르는 긴 봉/커튼봉 형태의 막대가 두 개 겹쳐 소파 위 공간을 관통하듯 그려져 방의 기하와 충돌한다.",
      "entities": "정인우: 20대 중반 한국 남성, 짧은 검은 머리, 감색 재킷 — 캐릭터 레퍼런스와 대체로 일치(비니는 없음). 고산들: 40대 남성 침입자, 어두운 재킷, 정인우가 아닌 인물이지만 샷 텍스트가 지정한 상대이므로 허용. 삼단봉: 실제로는 매우 긴 두 개의 금속 파이프로 렌더되어 '손에 쥔 삼단봉' 규격에 어긋남. 김지후는 오프스크린(지시대로).",
      "hard_violations": [
       "손에 쥔 삼단봉이 두 개의 긴 금속 막대로 중복 렌더되어 프레임을 X자로 가로지름 — 소품 중복/물리적으로 불가능한 파지"
      ],
      "physics": "정인우의 두 발은 바닥에 있고 팔은 어깨에서 올라가 지지는 성립. 그러나 아래로 뻗은 두 번째 막대는 어떤 손에도 잡혀 있지 않고 공중에서 시작해 오른쪽 아래로 관통하며, 무엇도 지지하지 않는다. 고산들은 두 발로 바닥을 딛고 있어 문제 없음."
     },
     {
      "label": "B",
      "direction": "정인우는 좌측 전경, 3/4 측면으로 서서 오른손에 쥔 삼단봉을 좌상단 대각선으로 높이 치켜들었다 — 봉의 방향이 상단 프레임으로 정확히 대각선을 그린다. 그의 몸과 팔은 우측의 고산들 진로를 가로막는 방향. 고산들은 중우측 중경에서 화면 좌측 아래(정인우 너머 오프스크린 김지후 쪽)로 시선을 던진다.",
      "built_space": "이전 스틸의 거실이 그대로: 좌측 현관벽·도어록·테이블보 씌운 식탁, 정면 발코니 창과 야경, 에어컨, 소파, 넘어진 액자와 리클라이너, 우측 책장과 바구니, 바닥의 깨진 화분·흙·유리 파편·핏자국. 구조물 중복 없음. 삼단봉 한 자루만 존재.",
      "entities": "정인우: 20대 중반 한국 남성, 단정한 검은 머리, 감색 재킷 안 짙은 니트 — 레퍼런스와 부합(비니 미착용은 이전 스틸에도 없던 항목이라 무해). 고산들: 침입자 중년 남성, 어두운 재킷·회색 티. 삼단봉: 검은 그립 + 은색 신축봉으로 실제 삼단봉의 형태·질감·무게감이 맞음. 김지후 오프스크린.",
      "hard_violations": [],
      "physics": "정인우는 두 발이 바닥에 있고 어깨-팔꿈치-손목이 이어진 자연스러운 상단 스윙 자세로 봉을 쥐고 있으며, 손이 그립을 정확히 감싸 무게를 지지한다. 고산들은 오른발 뒤·왼발 앞으로 전진 중인 체중 이동이 읽히고 두 발 모두 바닥에 접지. 공중에 뜬 물체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "동일 구도·인물·조명이지만 '손에 쥔 삼단봉' 하나가 두 개의 긴 금속 막대로 중복 렌더되어 프레임을 X자로 관통하고, 아래로 뻗은 막대는 어떤 손도 잡고 있지 않아 소품 중복·무지지 하드 위반."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "삼단봉 한 자루를 오른손으로 쥐고 좌상단 대각선으로 치켜든 순간이 정확하며, 좌전경 정인우·우중경 고산들 배치, 가슴 높이 3/4 측면 앙각, 이전 스틸의 파손 가구·핏자국·야간 저조도까지 계약대로 유지 — 인물·의상·배경을 바꾸지 않고 봉 중복만 해소한 이득이 명확."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "동일 구도·인물·조명이지만 '손에 쥔 삼단봉' 하나가 두 개의 긴 금속 막대로 중복 렌더되어 프레임을 X자로 관통하고, 아래로 뻗은 막대는 어떤 손도 잡고 있지 않아 소품 중복·무지지 하드 위반."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "삼단봉 한 자루를 오른손으로 쥐고 좌상단 대각선으로 치켜든 순간이 정확하며, 좌전경 정인우·우중경 고산들 배치, 가슴 높이 3/4 측면 앙각, 이전 스틸의 파손 가구·핏자국·야간 저조도까지 계약대로 유지 — 인물·의상·배경을 바꾸지 않고 봉 중복만 해소한 이득이 명확."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "정인우는 좌측 전경에서 우측(고산들 방향)을 향해 옆얼굴을 보이며 서 있고, 오른손은 위로 들려 있다. 그러나 손에서 뻗은 봉이 하나가 아니라 두 개의 긴 금속 막대로 보이며, 하나는 오른쪽 위 천장 쪽으로, 다른 하나는 오른쪽 아래(고산들 발치 쪽)로 X자를 그리며 프레임을 가로지른다. 고산들은 중우측 중경에 서서 화면 좌측(정인우/오프스크린 김지후 방향)으로 시선을 보내고 있어 시선 방향 자체는 맞다.",
      "built_space": "이전 스틸의 거실 구조(현관 쪽 벽·도어록, 좌측 테이블, 정면 발코니 창, 우측 소파와 넘어진 액자·리클라이너, 우측 책장, 바닥의 화분과 흙)가 대체로 유지된다. 다만 상단을 가로지르는 긴 봉/커튼봉 형태의 막대가 두 개 겹쳐 소파 위 공간을 관통하듯 그려져 방의 기하와 충돌한다.",
      "entities": "정인우: 20대 중반 한국 남성, 짧은 검은 머리, 감색 재킷 — 캐릭터 레퍼런스와 대체로 일치(비니는 없음). 고산들: 40대 남성 침입자, 어두운 재킷, 정인우가 아닌 인물이지만 샷 텍스트가 지정한 상대이므로 허용. 삼단봉: 실제로는 매우 긴 두 개의 금속 파이프로 렌더되어 '손에 쥔 삼단봉' 규격에 어긋남. 김지후는 오프스크린(지시대로).",
      "hard_violations": [
       "손에 쥔 삼단봉이 두 개의 긴 금속 막대로 중복 렌더되어 프레임을 X자로 가로지름 — 소품 중복/물리적으로 불가능한 파지"
      ],
      "physics": "정인우의 두 발은 바닥에 있고 팔은 어깨에서 올라가 지지는 성립. 그러나 아래로 뻗은 두 번째 막대는 어떤 손에도 잡혀 있지 않고 공중에서 시작해 오른쪽 아래로 관통하며, 무엇도 지지하지 않는다. 고산들은 두 발로 바닥을 딛고 있어 문제 없음."
     },
     {
      "label": "A",
      "direction": "정인우는 좌측 전경, 3/4 측면으로 서서 오른손에 쥔 삼단봉을 좌상단 대각선으로 높이 치켜들었다 — 봉의 방향이 상단 프레임으로 정확히 대각선을 그린다. 그의 몸과 팔은 우측의 고산들 진로를 가로막는 방향. 고산들은 중우측 중경에서 화면 좌측 아래(정인우 너머 오프스크린 김지후 쪽)로 시선을 던진다.",
      "built_space": "이전 스틸의 거실이 그대로: 좌측 현관벽·도어록·테이블보 씌운 식탁, 정면 발코니 창과 야경, 에어컨, 소파, 넘어진 액자와 리클라이너, 우측 책장과 바구니, 바닥의 깨진 화분·흙·유리 파편·핏자국. 구조물 중복 없음. 삼단봉 한 자루만 존재.",
      "entities": "정인우: 20대 중반 한국 남성, 단정한 검은 머리, 감색 재킷 안 짙은 니트 — 레퍼런스와 부합(비니 미착용은 이전 스틸에도 없던 항목이라 무해). 고산들: 침입자 중년 남성, 어두운 재킷·회색 티. 삼단봉: 검은 그립 + 은색 신축봉으로 실제 삼단봉의 형태·질감·무게감이 맞음. 김지후 오프스크린.",
      "hard_violations": [],
      "physics": "정인우는 두 발이 바닥에 있고 어깨-팔꿈치-손목이 이어진 자연스러운 상단 스윙 자세로 봉을 쥐고 있으며, 손이 그립을 정확히 감싸 무게를 지지한다. 고산들은 오른발 뒤·왼발 앞으로 전진 중인 체중 이동이 읽히고 두 발 모두 바닥에 접지. 공중에 뜬 물체 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 5
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S12sh2"
  }
 },
 "S12sh11": {
  "input_fingerprint": "c1bb7e4b10a83ed8",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 거실 바닥에 억눌린 채 공포에 질려 사색이 된 얼굴로 고산들을 올려다보는 김지후(강수리영)의 클로즈업.\n\nLOCATION (lock): The blood-marked floor of the wrecked living room inside the multi-family home. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the dolly-in near floor level beside 김지후(강수리영), angling upward along her eyeline so her terrified face occupies most of the close frame. 고산들 is retained only as a partial upper-edge torso and restraining arm, large enough to establish the source of pressure without displacing her as the primary subject; the emphasized change is the camera's floor-level distance to her face.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, foreground, looks toward 고산들 above her; 고산들 in the upper-right of the frame, foreground, looks toward 김지후 pinned below.\n- KEY BACKGROUND ELEMENTS: living-room floor (김지후 held against it); used as Supports 김지후's pinned body and fixes the camera at her vulnerable level; fragments of broken furniture (scattered nearby) — Small broken edges enter obliquely at the background margins; used as Appears at the frame margins to preserve continuity with the destroyed room; bloodstains (visible near the floor-level composition); used as Provides restrained visual evidence of the preceding violence behind the close confrontation.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient illumination appropriate to the nighttime interior isolates 김지후's pallid, terrified face while 고산들 remains a heavier partial mass above her.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same wrecked living-room environment, blood marks, and nighttime illumination. Exclude the standing confrontation and most furniture from the close-up; show the woman pinned on the floor looking upward.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears the white animal-bone necklace while pinned on the floor; the wreckage and bloodstains remain around her.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 9,
   "C": 7,
   "B": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 9,
    "verdict_ko": "바닥 높이에서 얼굴을 올려다보는 카메라 축, 얼굴이 화면 대부분을 차지하는 클로즈업, 우상단에 걸린 고산들의 부분 몸통·제압 팔, 뼈 목걸이와 혈흔·파편 연속성까지 지시를 가장 정확히 충족."
   },
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "시선·제압 관계와 장소 연속성은 좋으나 카메라가 바닥 높이가 아닌 부감에 가깝고 얼굴 크기가 지정된 클로즈업보다 작다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "감정 연기와 혈흔은 강하지만 프레이밍이 미디엄에 가깝게 넓어지고 카메라가 내려다보는 각도라 '바닥 높이 클로즈업' 지시에서 가장 멀며, 상단 남성 의상도 이전 스틸의 다른 인물처럼 읽힌다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S12sh6_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:904317>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "김지후의 캐릭터 레퍼런스에 포함된 파란색 작업 조끼가 누락되어 회색 티셔츠만 입고 있습니다.",
     "fix_en": "Render the blue utility vest with reflective stripes from Jihu's character reference over her grey t-shirt, ensuring the vest's shoulder straps and chest panels are visible. Keep Jihu's face, expression, hair, bone necklace, her pose on the floor, the restraining man, the background, the lighting, and the framing exactly the same.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "화면 하단에 억누르는 남성(갈색 재킷)의 팔 외에 녹색 소매를 입은 정체불명의 팔이 하나 더 튀어나와 해부학적 오류를 범하고 있습니다.",
     "fix_en": "Remove the disembodied arm with the green cuff at the bottom edge of the frame, replacing it with the wooden floor and strands of Jihu's black hair. Maintain Jihu's face, body, clothing, bone necklace, the single restraining man in the brown jacket at the top right, the background wreckage, bloodstains, lighting, and framing unchanged.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Render the blue utility vest with reflective stripes from Jihu's character reference over her grey t-shirt, ensuring the vest's shoulder straps and chest panels are visible. Keep Jihu's face, expression, hair, bone necklace, her pose on the floor, the restraining man, the background, the lighting, and the framing exactly the same.\n- Remove the disembodied arm with the green cuff at the bottom edge of the frame, replacing it with the wooden floor and strands of Jihu's black hair. Maintain Jihu's face, body, clothing, bone necklace, the single restraining man in the brown jacket at the top right, the background wreckage, bloodstains, lighting, and framing unchanged.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·시선·바닥 지지 모두 계약을 충족하고 제압 팔이 더 명확히 보이는 이점은 있으나, 이전 스틸/캐릭터 레퍼런스로 락된 파란 안전조끼가 사라져 회색 티셔츠만 남은 의상 이탈이 결정적 손실이다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 로우앵글 클로즈업·시선 교차·야간 저조도·혈흔과 파편 연속성을 유지하면서 레퍼런스가 락한 파란 안전조끼와 뼈 목걸이까지 복원해 의상 지속 조항을 지켰고, 배경·얼굴·조명 등 요청되지 않은 변경은 거의 없다(제압 팔이 덜 보이는 것이 유일한 대가)."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.75,
      "B": 1.75
     },
     "adjusted": {
      "A": 1.75,
      "B": 1.75
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "A"
     },
     "agreed": false
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·시선·바닥 지지 모두 계약을 충족하고 제압 팔이 더 명확히 보이는 이점은 있으나, 이전 스틸/캐릭터 레퍼런스로 락된 파란 안전조끼가 사라져 회색 티셔츠만 남은 의상 이탈이 결정적 손실이다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 로우앵글 클로즈업·시선 교차·야간 저조도·혈흔과 파편 연속성을 유지하면서 레퍼런스가 락한 파란 안전조끼와 뼈 목걸이까지 복원해 의상 지속 조항을 지켰고, 배경·얼굴·조명 등 요청되지 않은 변경은 거의 없다(제압 팔이 덜 보이는 것이 유일한 대가)."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 바닥에 옆으로 누운 채 시선이 프레임 위쪽(우상단의 남성 상반신)을 향해 올려다본다. 카메라는 바닥 높이에서 그녀의 얼굴을 올려다보듯 근접해 있고, 상단 우측의 고산들은 턱과 어깨, 팔만 부분적으로 들어와 아래의 그녀를 내려다보는 방향으로 기울어 있다. 시선 교차가 성립한다.",
      "built_space": "무너진 거실: 뒤집힌 의자 다리, 깨진 화분과 흩어진 흙·잎, 유리·나무 파편, 마룻바닥, 배경 창문의 야간 실외와 천장 조명 하나. 이전 스틸의 거실 재질(원목 마루, 흰 벽, 커튼/창)과 일치하며 가구는 대부분 배제되어 프레임 가장자리에만 파편이 들어온다.",
      "entities": "김지후: 한국인 20대 초반 여성, 긴 검은 머리, 창백한 얼굴 — 캐릭터 레퍼런스와 부합. 의상: 파란 반사띠 조끼 + 회색 티셔츠로 레퍼런스와 일치하며 조끼의 로고/반사 스트라이프까지 보인다. 흰 동물뼈 목걸이가 목 옆에 놓여 있어 CARRIED STATE 충족. 고산들: 어두운 재킷의 성인 남성 부분 상반신(턱, 어깨, 팔)로 이전 스틸의 검은 재킷과 일치.",
      "hard_violations": [],
      "physics": "몸은 마룻바닥에 완전히 접촉해 누워 있고 머리·어깨·머리카락이 바닥에 눌려 퍼져 있다. 위에서 누르는 남성의 몸무게가 화면 우상단에서 내려오며 압박의 근원을 설명한다. 공중에 떠 있는 물체 없음. 유리·나무 파편과 혈흔은 바닥면에 놓여 원근에 맞게 축소되어 있다."
     },
     {
      "label": "B",
      "direction": "A와 동일하게 김지후의 시선은 위쪽 남성을 향하고, 카메라는 바닥 높이에서 올려다본다. 고산들의 팔은 우측에서 내려와 그녀 쪽으로 향한다. 방향성 자체는 문제 없음.",
      "built_space": "배경은 A와 사실상 동일(뒤집힌 의자, 깨진 화분, 파편, 창). 다만 우측 하단에 고산들의 팔/소매가 크게 확대되어 들어와 김지후의 상체와 조끼를 대부분 가린다.",
      "entities": "김지후 얼굴·머리는 A와 동일. 그러나 의상이 바뀌었다 — 레퍼런스와 이전 상황이 고정한 파란 반사조끼가 사라지고 회색 상의만 보인다. 흰 뼈 목걸이는 목/가슴 위에 더 크고 여러 가닥으로 렌더되어 형태가 달라졌다. 고산들의 팔이 하나 더 늘어난 듯 우측 상단과 우측 하단 두 곳에 어두운 팔 덩어리가 각각 들어와 팔 개수/배치가 모호하다.",
      "hard_violations": [
       "의상 락 위반: 김지후의 파란 반사조끼가 소실되어 이전 스틸/캐릭터 레퍼런스가 고정한 복장이 바뀜"
      ],
      "physics": "몸은 바닥에 접촉해 있고 지지 문제는 없다. 다만 우측 하단의 큰 팔 덩어리는 어깨나 몸통으로 이어지는 경로가 불분명해 어느 신체에 붙었는지 읽히지 않는다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "바닥 높이 클로즈업, 시선 상향, 좌중앙 배치, 우상단 부분 상반신·제압 팔, 야간 저조도, 파편·혈흔 여백 배치까지 지시를 모두 충족하고 레퍼런스의 파란 반사조끼와 흰 뼈 목걸이도 그대로 유지했다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "구도·시선은 A와 같으나 계약이 요구하지 않은 변경으로 김지후의 파란 반사조끼가 사라지고 목걸이 형태가 달라졌으며, 우측 하단에 정체 불명의 큰 팔 덩어리가 추가되어 인물 의상 락과 신체 판독성을 함께 잃었다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "바닥 높이 클로즈업, 시선 상향, 좌중앙 배치, 우상단 부분 상반신·제압 팔, 야간 저조도, 파편·혈흔 여백 배치까지 지시를 모두 충족하고 레퍼런스의 파란 반사조끼와 흰 뼈 목걸이도 그대로 유지했다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "구도·시선은 A와 같으나 계약이 요구하지 않은 변경으로 김지후의 파란 반사조끼가 사라지고 목걸이 형태가 달라졌으며, 우측 하단에 정체 불명의 큰 팔 덩어리가 추가되어 인물 의상 락과 신체 판독성을 함께 잃었다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "김지후는 바닥에 옆으로 누운 채 시선이 프레임 위쪽(우상단의 남성 상반신)을 향해 올려다본다. 카메라는 바닥 높이에서 그녀의 얼굴을 올려다보듯 근접해 있고, 상단 우측의 고산들은 턱과 어깨, 팔만 부분적으로 들어와 아래의 그녀를 내려다보는 방향으로 기울어 있다. 시선 교차가 성립한다.",
      "built_space": "무너진 거실: 뒤집힌 의자 다리, 깨진 화분과 흩어진 흙·잎, 유리·나무 파편, 마룻바닥, 배경 창문의 야간 실외와 천장 조명 하나. 이전 스틸의 거실 재질(원목 마루, 흰 벽, 커튼/창)과 일치하며 가구는 대부분 배제되어 프레임 가장자리에만 파편이 들어온다.",
      "entities": "김지후: 한국인 20대 초반 여성, 긴 검은 머리, 창백한 얼굴 — 캐릭터 레퍼런스와 부합. 의상: 파란 반사띠 조끼 + 회색 티셔츠로 레퍼런스와 일치하며 조끼의 로고/반사 스트라이프까지 보인다. 흰 동물뼈 목걸이가 목 옆에 놓여 있어 CARRIED STATE 충족. 고산들: 어두운 재킷의 성인 남성 부분 상반신(턱, 어깨, 팔)로 이전 스틸의 검은 재킷과 일치.",
      "hard_violations": [],
      "physics": "몸은 마룻바닥에 완전히 접촉해 누워 있고 머리·어깨·머리카락이 바닥에 눌려 퍼져 있다. 위에서 누르는 남성의 몸무게가 화면 우상단에서 내려오며 압박의 근원을 설명한다. 공중에 떠 있는 물체 없음. 유리·나무 파편과 혈흔은 바닥면에 놓여 원근에 맞게 축소되어 있다."
     },
     {
      "label": "A",
      "direction": "A와 동일하게 김지후의 시선은 위쪽 남성을 향하고, 카메라는 바닥 높이에서 올려다본다. 고산들의 팔은 우측에서 내려와 그녀 쪽으로 향한다. 방향성 자체는 문제 없음.",
      "built_space": "배경은 A와 사실상 동일(뒤집힌 의자, 깨진 화분, 파편, 창). 다만 우측 하단에 고산들의 팔/소매가 크게 확대되어 들어와 김지후의 상체와 조끼를 대부분 가린다.",
      "entities": "김지후 얼굴·머리는 A와 동일. 그러나 의상이 바뀌었다 — 레퍼런스와 이전 상황이 고정한 파란 반사조끼가 사라지고 회색 상의만 보인다. 흰 뼈 목걸이는 목/가슴 위에 더 크고 여러 가닥으로 렌더되어 형태가 달라졌다. 고산들의 팔이 하나 더 늘어난 듯 우측 상단과 우측 하단 두 곳에 어두운 팔 덩어리가 각각 들어와 팔 개수/배치가 모호하다.",
      "hard_violations": [
       "의상 락 위반: 김지후의 파란 반사조끼가 소실되어 이전 스틸/캐릭터 레퍼런스가 고정한 복장이 바뀜"
      ],
      "physics": "몸은 바닥에 접촉해 있고 지지 문제는 없다. 다만 우측 하단의 큰 팔 덩어리는 어깨나 몸통으로 이어지는 경로가 불분명해 어느 신체에 붙었는지 읽히지 않는다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 10,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S12sh6"
  }
 },
 "S16sh1::bgfirst_bg": {
  "input_fingerprint": "1e2c1fb348b6e675",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 컴퓨터 모니터 화면에 얼굴을 바짝 들이댄 정인우의 측면.\n\nLOCATION (lock): A CCTV workstation within the police detective office, lit by the computer monitor around dawn.\n\nTIME OF DAY (lock): dawn.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At In-woo's seated eye level, finish the dolly-in beside him so his tense profile and the monitor share a close lateral composition, with his face on the left and the CCTV image on the right. He hunches toward the screen until the furrow of his brow nearly meets its edge, while the sling-supported arm remains a partial lower-frame reminder of the previous scene.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: computer monitor (Displaying CCTV footage) — Its front screen faces partly toward the camera and shows CCTV footage of Sandeul carrying the unconscious Jihu out of the house; used as Shares focus with In-woo's profile and provides the visual evidence driving his hostility; workstation (In use) — Its user-facing side aligns with In-woo's seated profile; used as Anchors In-woo's investigative position before the camera later moves behind him.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The detective-office setting remains dark as specified around the monitor, with low-key contrast and restrained neutral color emphasizing In-woo's hostile concentration.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 컴퓨터 모니터 화면에 얼굴을 바짝 들이댄 정인우의 측면.\n\nLOCATION (lock): A CCTV workstation within the police detective office, lit by the computer monitor around dawn.\n\nTIME OF DAY (lock): dawn.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At In-woo's seated eye level, finish the dolly-in beside him so his tense profile and the monitor share a close lateral composition, with his face on the left and the CCTV image on the right. He hunches toward the screen until the furrow of his brow nearly meets its edge, while the sling-supported arm remains a partial lower-frame reminder of the previous scene.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: computer monitor (Displaying CCTV footage) — Its front screen faces partly toward the camera and shows CCTV footage of Sandeul carrying the unconscious Jihu out of the house; used as Shares focus with In-woo's profile and provides the visual evidence driving his hostility; workstation (In use) — Its user-facing side aligns with In-woo's seated profile; used as Anchors In-woo's investigative position before the camera later moves behind him.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The detective-office setting remains dark as specified around the monitor, with low-key contrast and restrained neutral color emphasizing In-woo's hostile concentration.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S16sh1__bgfirst_bg.png",
  "asset_id": "b058d2be-d53b-477b-8f03-e39b6d546508",
  "input_asset_ids": [
   "f93fe4e8-3cd3-4e25-a5da-16b9fb257ecb",
   "c86e9107-089f-4612-9fc8-18a83b823920"
  ]
 },
 "S16sh1": {
  "input_fingerprint": "1740f7bf458471f8",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 어두운 컴퓨터 모니터 화면에 얼굴을 바짝 들이댄 정인우의 측면.\n\nLOCATION (lock): A CCTV workstation within the police detective office, lit by the computer monitor around dawn. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At In-woo's seated eye level, finish the dolly-in beside him so his tense profile and the monitor share a close lateral composition, with his face on the left and the CCTV image on the right. He hunches toward the screen until the furrow of his brow nearly meets its edge, while the sling-supported arm remains a partial lower-frame reminder of the previous scene.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: computer monitor (Displaying CCTV footage) — Its front screen faces partly toward the camera and shows CCTV footage of Sandeul carrying the unconscious Jihu out of the house; used as Shares focus with In-woo's profile and provides the visual evidence driving his hostility; workstation (In use) — Its user-facing side aligns with In-woo's seated profile; used as Anchors In-woo's investigative position before the camera later moves behind him.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The detective-office setting remains dark as specified around the monitor, with low-key contrast and restrained neutral color emphasizing In-woo's hostile concentration.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains fixed in the same sling while he examines the CCTV footage.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 어두운 컴퓨터 모니터 화면에 얼굴을 바짝 들이댄 정인우의 측면.\n\nLOCATION (lock): A CCTV workstation within the police detective office, lit by the computer monitor around dawn. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At In-woo's seated eye level, finish the dolly-in beside him so his tense profile and the monitor share a close lateral composition, with his face on the left and the CCTV image on the right. He hunches toward the screen until the furrow of his brow nearly meets its edge, while the sling-supported arm remains a partial lower-frame reminder of the previous scene.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: computer monitor (Displaying CCTV footage) — Its front screen faces partly toward the camera and shows CCTV footage of Sandeul carrying the unconscious Jihu out of the house; used as Shares focus with In-woo's profile and provides the visual evidence driving his hostility; workstation (In use) — Its user-facing side aligns with In-woo's seated profile; used as Anchors In-woo's investigative position before the camera later moves behind him.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The detective-office setting remains dark as specified around the monitor, with low-key contrast and restrained neutral color emphasizing In-woo's hostile concentration.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains fixed in the same sling while he examines the CCTV footage.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 어두운 컴퓨터 모니터 화면에 얼굴을 바짝 들이댄 정인우의 측면.\n\nLOCATION (lock): A CCTV workstation within the police detective office, lit by the computer monitor around dawn. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At In-woo's seated eye level, finish the dolly-in beside him so his tense profile and the monitor share a close lateral composition, with his face on the left and the CCTV image on the right. He hunches toward the screen until the furrow of his brow nearly meets its edge, while the sling-supported arm remains a partial lower-frame reminder of the previous scene.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: computer monitor (Displaying CCTV footage) — Its front screen faces partly toward the camera and shows CCTV footage of Sandeul carrying the unconscious Jihu out of the house; used as Shares focus with In-woo's profile and provides the visual evidence driving his hostility; workstation (In use) — Its user-facing side aligns with In-woo's seated profile; used as Anchors In-woo's investigative position before the camera later moves behind him.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The detective-office setting remains dark as specified around the monitor, with low-key contrast and restrained neutral color emphasizing In-woo's hostile concentration.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains fixed in the same sling while he examines the CCTV footage.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S16sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S16sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:871658>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L14B03.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:871658>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "all_candidates_fail": false,
    "ranking": [
     "A",
     "B"
    ],
    "readings": [
     {
      "built_space": "경찰 사무실의 사용 중인 CCTV 워크스테이션으로 보이며, 전면이 인우와 카메라 쪽으로 비스듬히 향한 모니터 1대와 책상, 모니터 받침이 보인다. 인우는 화면 왼쪽의 좌석 위치에서 사용자 방향으로 몸을 숙이고 있다. 뒤쪽의 검은 다중 모니터 프레임과 수납장은 장소 사진의 CCTV 설비와 잘 이어진다.",
      "direction": "인우의 얼굴과 두 눈은 오른쪽 모니터의 CCTV 영상, 특히 사람을 안고 있는 남성을 향한다. 얼굴이 화면 가장자리 가까이 접근해 있으며 시선이 명확히 화면에 닿는다. CCTV 속 운반자는 안은 인물을 지지한 채 복도 방향으로 이동하는 모습이다.",
      "entities": "왼쪽에는 검은 머리의 젊은 한국 남성 정인우가 긴장된 측면 얼굴로 보이고, 아래에는 부상당한 팔을 고정한 짙은 남색 슬링이 분명히 보인다. 오른쪽 모니터에는 산들이 의식 없는 지후를 두 팔로 안고 옮기는 CCTV 영상이 있다. 인우의 얼굴은 참고 인물과 대체로 부합하지만 흰 셔츠 차림은 캐릭터 참고의 재킷·니트와 다르다.",
      "hard_violations": [],
      "label": "A",
      "physics": "인우의 상체는 좌석과 책상 앞 자세로 지지되며, 슬링은 목의 스트랩과 몸 앞의 포켓으로 부상당한 팔을 받친다. CCTV 속 지후는 산들의 한 팔이 등과 어깨를, 다른 팔이 무릎 아래를 받치는 크래들 캐리 자세여서 공중에 뜬 신체의 지지가 명확하다."
     },
     {
      "built_space": "모니터 1대가 워크스테이션 위에 있고 사용자 쪽 화면이 인우와 카메라에 비스듬히 노출된다. 인우는 화면 왼쪽 좌석에서 모니터 쪽으로 숙였으며, 책상 위 문구류와 뒤쪽 시계·서류 설비가 경찰 사무 공간을 이룬다. 다만 장소 사진의 특징적인 검은 다중 CCTV 랙 구조는 A보다 덜 드러난다.",
      "direction": "인우의 눈과 찌푸린 얼굴은 오른쪽 모니터 속 두 사람에게 정확히 향한다. CCTV 속 산들은 의식 없는 지후를 안고 열린 출입구를 통과해 바깥으로 나가는 방향이라 ‘집에서 데리고 나오는’ 행동이 선명하다.",
      "entities": "왼쪽에는 젊은 한국 남성 정인우의 측면 얼굴이 크게 보이고, 참고와 유사한 어두운 재킷과 녹색 상의가 확인된다. 하단에는 슬링 일부와 자유로운 반대쪽 손이 보인다. 모니터에는 산들이 지후를 안고 나오는 모습이 표시된다. 그러나 CCTV 화면 상단에 시간 정보 형태의 문자와 숫자가 노출되어 있다.",
      "hard_violations": [
       "화면 상단에 CCTV 날짜·시간 형태의 문자와 숫자가 노출되어 ‘텍스트가 없어야 한다’는 명시 조건을 위반한다."
      ],
      "label": "B",
      "physics": "인우는 좌석 앞 책상에 몸을 기울이고 자유로운 손을 책상에 둔 자세이며, 하단의 슬링이 다친 팔을 받친다. CCTV 속 지후는 산들의 두 팔에 의해 등과 다리 쪽이 지지되어 있고, 산들의 다리도 바닥을 딛고 있어 운반 동작이 물리적으로 성립한다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 9,
      "verdict_ko": "텍스트 누출 없이 인우의 긴장된 측면, 화면을 향한 정확한 시선, 슬링과 CCTV 운반 장면을 요구된 좌우 근접 구도로 가장 충실히 구현했다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "더 밀착된 클로즈업과 집 밖으로 운반하는 CCTV 동작은 뛰어나지만, 화면의 날짜·시간 문자가 명시적인 무텍스트 조건을 위반해 결정적으로 뒤처진다."
     }
    ],
    "winner": "A"
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 9,
      "verdict_ko": "텍스트 누출 없이 인우의 긴장된 측면, 화면을 향한 정확한 시선, 슬링과 CCTV 운반 장면을 요구된 좌우 근접 구도로 가장 충실히 구현했다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "더 밀착된 클로즈업과 집 밖으로 운반하는 CCTV 동작은 뛰어나지만, 화면의 날짜·시간 문자가 명시적인 무텍스트 조건을 위반해 결정적으로 뒤처진다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "built_space": "경찰 사무실의 사용 중인 CCTV 워크스테이션으로 보이며, 전면이 인우와 카메라 쪽으로 비스듬히 향한 모니터 1대와 책상, 모니터 받침이 보인다. 인우는 화면 왼쪽의 좌석 위치에서 사용자 방향으로 몸을 숙이고 있다. 뒤쪽의 검은 다중 모니터 프레임과 수납장은 장소 사진의 CCTV 설비와 잘 이어진다.",
      "direction": "인우의 얼굴과 두 눈은 오른쪽 모니터의 CCTV 영상, 특히 사람을 안고 있는 남성을 향한다. 얼굴이 화면 가장자리 가까이 접근해 있으며 시선이 명확히 화면에 닿는다. CCTV 속 운반자는 안은 인물을 지지한 채 복도 방향으로 이동하는 모습이다.",
      "entities": "왼쪽에는 검은 머리의 젊은 한국 남성 정인우가 긴장된 측면 얼굴로 보이고, 아래에는 부상당한 팔을 고정한 짙은 남색 슬링이 분명히 보인다. 오른쪽 모니터에는 산들이 의식 없는 지후를 두 팔로 안고 옮기는 CCTV 영상이 있다. 인우의 얼굴은 참고 인물과 대체로 부합하지만 흰 셔츠 차림은 캐릭터 참고의 재킷·니트와 다르다.",
      "hard_violations": [],
      "label": "A",
      "physics": "인우의 상체는 좌석과 책상 앞 자세로 지지되며, 슬링은 목의 스트랩과 몸 앞의 포켓으로 부상당한 팔을 받친다. CCTV 속 지후는 산들의 한 팔이 등과 어깨를, 다른 팔이 무릎 아래를 받치는 크래들 캐리 자세여서 공중에 뜬 신체의 지지가 명확하다."
     },
     {
      "built_space": "모니터 1대가 워크스테이션 위에 있고 사용자 쪽 화면이 인우와 카메라에 비스듬히 노출된다. 인우는 화면 왼쪽 좌석에서 모니터 쪽으로 숙였으며, 책상 위 문구류와 뒤쪽 시계·서류 설비가 경찰 사무 공간을 이룬다. 다만 장소 사진의 특징적인 검은 다중 CCTV 랙 구조는 A보다 덜 드러난다.",
      "direction": "인우의 눈과 찌푸린 얼굴은 오른쪽 모니터 속 두 사람에게 정확히 향한다. CCTV 속 산들은 의식 없는 지후를 안고 열린 출입구를 통과해 바깥으로 나가는 방향이라 ‘집에서 데리고 나오는’ 행동이 선명하다.",
      "entities": "왼쪽에는 젊은 한국 남성 정인우의 측면 얼굴이 크게 보이고, 참고와 유사한 어두운 재킷과 녹색 상의가 확인된다. 하단에는 슬링 일부와 자유로운 반대쪽 손이 보인다. 모니터에는 산들이 지후를 안고 나오는 모습이 표시된다. 그러나 CCTV 화면 상단에 시간 정보 형태의 문자와 숫자가 노출되어 있다.",
      "hard_violations": [
       "화면 상단에 CCTV 날짜·시간 형태의 문자와 숫자가 노출되어 ‘텍스트가 없어야 한다’는 명시 조건을 위반한다."
      ],
      "label": "B",
      "physics": "인우는 좌석 앞 책상에 몸을 기울이고 자유로운 손을 책상에 둔 자세이며, 하단의 슬링이 다친 팔을 받친다. CCTV 속 지후는 산들의 두 팔에 의해 등과 다리 쪽이 지지되어 있고, 산들의 다리도 바닥을 딛고 있어 운반 동작이 물리적으로 성립한다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "얼굴 좌·모니터 우의 근접 측면 구도와 시선·화면 내용은 잘 맞으나, 하단의 슬링이 형태 불명확해 '팔걸이 유지' 계약이 약하고 배경이 새벽 저조도라기엔 밝고 CCTV 워크스테이션 느낌이 옅다.  ★위반: [gpt] CCTV 화면 상단에 날짜·시간 형태의 문자 오버레이가 노출되어 ‘어디에도 텍스트 없음’ 조건을 위반한다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "눈높이 측면 클로즈업, 얼굴 좌·CCTV 화면 우 배치, 화면 속 '의식 없는 지후를 안고 나가는 산들' 영상, 파란 슬링 유지, 새벽 저조도 모니터광까지 모두 충족해 지시 이행도가 가장 높다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.349,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.099,
      "B": 2.0
     },
     "violations": {
      "A": [
       "[gpt] CCTV 화면 상단에 날짜·시간 형태의 문자 오버레이가 노출되어 ‘어디에도 텍스트 없음’ 조건을 위반한다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "얼굴 좌·모니터 우의 근접 측면 구도와 시선·화면 내용은 잘 맞으나, 하단의 슬링이 형태 불명확해 '팔걸이 유지' 계약이 약하고 배경이 새벽 저조도라기엔 밝고 CCTV 워크스테이션 느낌이 옅다.  ★위반: [gpt] CCTV 화면 상단에 날짜·시간 형태의 문자 오버레이가 노출되어 ‘어디에도 텍스트 없음’ 조건을 위반한다."
     },
     {
      "label": "A",
      "score": 9,
      "verdict_ko": "눈높이 측면 클로즈업, 얼굴 좌·CCTV 화면 우 배치, 화면 속 '의식 없는 지후를 안고 나가는 산들' 영상, 파란 슬링 유지, 새벽 저조도 모니터광까지 모두 충족해 지시 이행도가 가장 높다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 18,
     "B": 12
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "readings": [
   {
    "built_space": "경찰 사무실의 사용 중인 CCTV 워크스테이션으로 보이며, 전면이 인우와 카메라 쪽으로 비스듬히 향한 모니터 1대와 책상, 모니터 받침이 보인다. 인우는 화면 왼쪽의 좌석 위치에서 사용자 방향으로 몸을 숙이고 있다. 뒤쪽의 검은 다중 모니터 프레임과 수납장은 장소 사진의 CCTV 설비와 잘 이어진다.",
    "direction": "인우의 얼굴과 두 눈은 오른쪽 모니터의 CCTV 영상, 특히 사람을 안고 있는 남성을 향한다. 얼굴이 화면 가장자리 가까이 접근해 있으며 시선이 명확히 화면에 닿는다. CCTV 속 운반자는 안은 인물을 지지한 채 복도 방향으로 이동하는 모습이다.",
    "entities": "왼쪽에는 검은 머리의 젊은 한국 남성 정인우가 긴장된 측면 얼굴로 보이고, 아래에는 부상당한 팔을 고정한 짙은 남색 슬링이 분명히 보인다. 오른쪽 모니터에는 산들이 의식 없는 지후를 두 팔로 안고 옮기는 CCTV 영상이 있다. 인우의 얼굴은 참고 인물과 대체로 부합하지만 흰 셔츠 차림은 캐릭터 참고의 재킷·니트와 다르다.",
    "hard_violations": [],
    "label": "A",
    "physics": "인우의 상체는 좌석과 책상 앞 자세로 지지되며, 슬링은 목의 스트랩과 몸 앞의 포켓으로 부상당한 팔을 받친다. CCTV 속 지후는 산들의 한 팔이 등과 어깨를, 다른 팔이 무릎 아래를 받치는 크래들 캐리 자세여서 공중에 뜬 신체의 지지가 명확하다."
   },
   {
    "built_space": "모니터 1대가 워크스테이션 위에 있고 사용자 쪽 화면이 인우와 카메라에 비스듬히 노출된다. 인우는 화면 왼쪽 좌석에서 모니터 쪽으로 숙였으며, 책상 위 문구류와 뒤쪽 시계·서류 설비가 경찰 사무 공간을 이룬다. 다만 장소 사진의 특징적인 검은 다중 CCTV 랙 구조는 A보다 덜 드러난다.",
    "direction": "인우의 눈과 찌푸린 얼굴은 오른쪽 모니터 속 두 사람에게 정확히 향한다. CCTV 속 산들은 의식 없는 지후를 안고 열린 출입구를 통과해 바깥으로 나가는 방향이라 ‘집에서 데리고 나오는’ 행동이 선명하다.",
    "entities": "왼쪽에는 젊은 한국 남성 정인우의 측면 얼굴이 크게 보이고, 참고와 유사한 어두운 재킷과 녹색 상의가 확인된다. 하단에는 슬링 일부와 자유로운 반대쪽 손이 보인다. 모니터에는 산들이 지후를 안고 나오는 모습이 표시된다. 그러나 CCTV 화면 상단에 시간 정보 형태의 문자와 숫자가 노출되어 있다.",
    "hard_violations": [
     "화면 상단에 CCTV 날짜·시간 형태의 문자와 숫자가 노출되어 ‘텍스트가 없어야 한다’는 명시 조건을 위반한다."
    ],
    "label": "B",
    "physics": "인우는 좌석 앞 책상에 몸을 기울이고 자유로운 손을 책상에 둔 자세이며, 하단의 슬링이 다친 팔을 받친다. CCTV 속 지후는 산들의 두 팔에 의해 등과 다리 쪽이 지지되어 있고, 산들의 다리도 바닥을 딛고 있어 운반 동작이 물리적으로 성립한다."
   }
  ],
  "totals": {
   "A": 18,
   "B": 12
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 9,
    "verdict_ko": "텍스트 누출 없이 인우의 긴장된 측면, 화면을 향한 정확한 시선, 슬링과 CCTV 운반 장면을 요구된 좌우 근접 구도로 가장 충실히 구현했다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "더 밀착된 클로즈업과 집 밖으로 운반하는 CCTV 동작은 뛰어나지만, 화면의 날짜·시간 문자가 명시적인 무텍스트 조건을 위반해 결정적으로 뒤처진다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L14B03.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:871658>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "화면 하단 슬링 바깥으로 노출된 오른손의 해부학적 형태가 완전히 뭉개져 살덩이처럼 보입니다.",
     "fix_en": "Redraw the right hand emerging from the white cast at the bottom of the frame to be an anatomically correct right hand resting naturally, or conceal the fingers completely inside the cast. Preserve In-woo's face, his white shirt, the blue sling straps, the computer monitor displaying the CCTV footage, and the entire room background exactly as they are.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the right hand emerging from the white cast at the bottom of the frame to be an anatomically correct right hand resting naturally, or conceal the fingers completely inside the cast. Preserve In-woo's face, his white shirt, the blue sling straps, the computer monitor displaying the CCTV footage, and the entire room background exactly as they are.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "시선이 모니터에 정확히 닿는 측면 클로즈업, 슬링 고정 팔이 하단 프레임에 남는 CARRIED STATE 충족, 얼굴이 화면 왼쪽·CCTV가 오른쪽이라는 구도 지시를 더 밀착된 클로즈업으로 실현했고 손·팔 해부가 깔끔하다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "동일 구도지만 카운터 위에 새 맨손을 추가해 슬링 팔과의 좌우 관계가 모호해지고 어깨-팔 연결이 어색해졌으며, 인물이 뒤로 물러나 '얼굴을 바짝 들이댄' 밀착감도 약화되어 계약이 요구하지 않은 손실이 더 크다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.625
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.625
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "시선이 모니터에 정확히 닿는 측면 클로즈업, 슬링 고정 팔이 하단 프레임에 남는 CARRIED STATE 충족, 얼굴이 화면 왼쪽·CCTV가 오른쪽이라는 구도 지시를 더 밀착된 클로즈업으로 실현했고 손·팔 해부가 깔끔하다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "동일 구도지만 카운터 위에 새 맨손을 추가해 슬링 팔과의 좌우 관계가 모호해지고 어깨-팔 연결이 어색해졌으며, 인물이 뒤로 물러나 '얼굴을 바짝 들이댄' 밀착감도 약화되어 계약이 요구하지 않은 손실이 더 크다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "클로즈업 스케일·좌측 프로필/우측 모니터 배치가 계약에 가장 근접하고, 카운터를 짚은 손과 슬링이 지지·부상 상태를 명확히 보여준다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "동일 구도지만 프레임이 넓어져 얼굴 크기가 줄어 '얼굴을 바짝 들이댄 클로즈업' 요구에서 후퇴했고, 계약이 요구하지 않은 손(붕대 감긴 오른손)의 제거·모호화라는 손실이 있다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.875,
      "B": 1.75
     },
     "adjusted": {
      "A": 1.875,
      "B": 1.75
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "B"
     },
     "agreed": false
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "클로즈업 스케일·좌측 프로필/우측 모니터 배치가 계약에 가장 근접하고, 카운터를 짚은 손과 슬링이 지지·부상 상태를 명확히 보여준다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "동일 구도지만 프레임이 넓어져 얼굴 크기가 줄어 '얼굴을 바짝 들이댄 클로즈업' 요구에서 후퇴했고, 계약이 요구하지 않은 손(붕대 감긴 오른손)의 제거·모호화라는 손실이 있다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 14,
     "B": 14
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S16sh1__bgfirst_bg.png",
   "bg_asset_id": "b058d2be-d53b-477b-8f03-e39b6d546508",
   "bg_record_key": "S16sh1::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S16sh5": {
  "input_fingerprint": "803b63b3edd90c67",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 핸드폰 화면을 내려다보며 도저히 이해할 수 없다는 듯 얼굴이 잔뜩 일그러진 정인우의 굳은 표정.\n\nLOCATION (lock): The desk area of the police detective office beside the active CCTV computer. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: End the upward tilt in a close front three-quarter view of In-woo, holding his rigid, baffled expression across the upper two-thirds while the phone remains legible at the lower edge. His torso pulls slightly away from the message, and the composition leaves lateral space in the direction he will turn toward Detective Noh in the next movement.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: mobile phone (Message displayed) — The screen faces up toward In-woo and partly toward the camera, displaying Jihu's message about Geumwoldo and her mother; used as Retains the cause of In-woo's reaction along the lower frame boundary; computer monitor (On with CCTV files open) — Only part of the active screen face remains visible behind In-woo; used as Preserves continuity with the CCTV investigation while remaining out of primary focus.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Subdued office illumination keeps the palette neutral and low-key, concentrating contrast on In-woo's tightened features and the active phone screen.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same detective-office desks, monitor glow, fluorescent ambience, and the injured man's sling. Exclude the CCTV-monitor close inspection; show him reacting to the phone message.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo continues to wear the sling on his injured arm.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 정인우 right now, so 정인우's hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 정인우: its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "인우의 시선은 화면(핸드폰)이 아니라 프레임 오른쪽 아래~카메라 쪽 허공을 향해 있어 '핸드폰 화면을 내려다보는' 지시와 어긋난다. 폰은 오른손에 들려 화면이 카메라 정면을 향하고 인우 눈 쪽으로는 향하지 않아 소품 방향 규칙(기능면은 사용자 눈으로) 위반에 가깝다. 폰은 하단 가장자리가 아니라 프레임 중앙 우측에 크게 놓여 있다.",
    "built_space": "형사과 사무실: 왼쪽 창, 벽시계, 서류철, 오른쪽에 CCTV 모니터. 이전 스틸의 모니터 영상(남자가 여자를 안고 있는 CCTV)이 그대로 재현되어 장소 연속성은 좋다. 다만 모니터가 배경에서 여전히 큰 전경급 존재감으로 남고, '일부만 보인다'는 지시보다 과하게 보인다.",
    "entities": "인우: 20대 중반이라기엔 다소 나이 들어 보이나 이전 스틸 인물과 동일 계열, 흰 셔츠+남색 팔걸이 슬링 유지. 폰: 검은 아이폰류로 프롭 레퍼런스와 유사. 그러나 화면에 영문 잡문자(가독 불가한 텍스트 덩어리)가 렌더되어 'No text' 규정 및 자연스러움을 해친다. 왼쪽 배경에 이전 스틸에 없던 다른 인물의 어깨/머리가 흐리게 추가돼 있다.",
    "hard_violations": [
     "샷 텍스트가 지정하지 않은 추가 인물(왼쪽 배경 흐린 인물)이 등장",
     "화면에 무의미한 영문 텍스트 덩어리가 렌더되어 텍스트 금지 위반"
    ],
    "physics": "폰은 오른손에 확실히 쥐어져 있고 팔꿈치가 자연스럽게 굽어 있다. 슬링에 든 왼팔이 앞에 있고, 몸은 책상 앞에 앉은 자세로 지지된다. 부유하는 물체는 없다."
   },
   {
    "label": "B",
    "direction": "인우의 시선은 폰이 아니라 카메라 왼쪽 전방(정면 가까이)을 향해 있어 '화면을 내려다보는' 지시와 어긋난다. 폰 화면은 인우 쪽이 아니라 카메라를 향해 완전히 열려 있어 소품 방향 규칙 위반. 폰 위치는 프레임 우측 중앙~하단.",
    "built_space": "사무실 배경은 형광등, 창, 벽시계, 우측 CCTV 모니터로 이전 스틸과 대체로 일치. 다만 모니터가 4분할 CCTV 화면으로 바뀌어 이전 스틸의 단일 복도 영상과 다르고, 화면 안에 'CCTV' 라벨 텍스트가 찍혀 있다.",
    "entities": "인우: 흰 셔츠+남색 슬링 유지, 얼굴은 레퍼런스 인물과 유사하나 표정이 '일그러짐'보다는 멍한 응시. 폰: 검정 스마트폰으로 프롭과 부합. 화면에 한국어 카톡풍 대화가 렌더되어 있어 텍스트 금지 조항과 충돌.",
    "hard_violations": [
     "모니터의 'CCTV' 라벨 및 폰 화면의 판독 가능한 텍스트가 프레임에 삽입되어 텍스트 금지 위반"
    ],
    "physics": "폰은 오른손 손가락에 쥐어져 있고 손목·전완이 프레임에 들어와 있다. 상체는 책상 쪽으로 기운 앉은 자세로 지지된다. 부유 물체 없음."
   },
   {
    "label": "C",
    "direction": "인우의 시선이 명확히 아래쪽 우측의 핸드폰 화면으로 내려가 있고, 눈썹이 찌푸려진 굳은 표정이다. 폰은 화면이 인우의 눈 쪽(위·안쪽)으로 향하고 카메라에는 등면(카메라 렌즈 모듈)이 보여 소품 방향 규칙과 정확히 일치. 몸통은 메시지에서 살짝 뒤로 물러나 있고 왼쪽에 여백이 남아 노 형사 쪽으로 돌 여지를 만든다.",
    "built_space": "형사과 사무실: 형광등, 창문, 서류철, 보온병, 우측 프레임 가장자리에 CCTV 모니터의 화면면 일부만 보임 — '활성 화면의 일부만 보인다'는 지시와 정확히 부합. 배경 요소가 전경으로 과장되지 않았다.",
    "entities": "인우: 흰 셔츠, 남색 팔걸이 슬링(붕대 포함) 유지, 이전 스틸 인물과 동일한 얼굴·헤어. 폰: 검정 바디에 후면 트리플 카메라 모듈이 보이는 아이폰류로 프롭 레퍼런스와 일치. 프레임 내 판독 가능한 텍스트·자막 없음. 추가 인물 없음.",
    "hard_violations": [],
    "physics": "폰은 오른손 손가락 네 개와 엄지로 확실히 쥐어져 있고 전완이 프레임 안에 들어 있다. 왼팔은 슬링에 지지, 상체는 책상 앞 앉은 자세. 부유하거나 지지 없는 물체 없음."
   }
  ],
  "totals": {
   "A": 3,
   "B": 4,
   "C": 8
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "시선이 폰을 벗어나고 화면을 카메라로 뒤집어 소품 방향을 위반했으며, 배경에 없던 인물과 화면 잡텍스트가 추가되어 하드 위반."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "장소·슬링 연속성은 무난하나 시선이 폰이 아닌 정면을 향하고 화면이 카메라로 열려 있으며, 모니터 'CCTV' 라벨 등 텍스트 삽입으로 감점."
   },
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "핸드폰 화면을 내려다보는 굳은 표정, 폰 후면이 카메라를 향하는 올바른 소품 방향, 모니터 일부만 남긴 배경, 슬링 유지까지 샷 텍스트와 프레이밍 지시를 가장 정확히 실현."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S16sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:871658>"
   },
   {
    "label": "PROP REFERENCE — 스마트폰: the exact object appearing in this shot; match its look, material and wear exactly.",
    "path": "<bytes:813831>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트에 지시된 바와 달리 핸드폰 화면과 메시지가 보이지 않고 카메라 렌즈가 있는 뒷면이 향해 있으며, 폰을 쥔 왼쪽 팔에 해부학적으로 잘못된 오른손(엄지가 카메라 기준 왼쪽에 위치)이 달려 있습니다.",
     "fix_en": "Redraw the mobile phone and the hand holding it so that a correctly anatomical left hand holds the device, and the phone's glowing screen faces partly toward the camera displaying a text message conversation, preserving the person present and his position, his clothing, the set, the light, and the framing exactly.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "이전 샷에서 왼쪽 어깨로 넘어가던 슬링 끈이 오른쪽 어깨 위로 잘못 걸쳐져 있어 의상 연속성에 어긋납니다.",
     "fix_en": "",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the mobile phone and the hand holding it so that a correctly anatomical left hand holds the device, and the phone's glowing screen faces partly toward the camera displaying a text message conversation, preserving the person present and his position, his clothing, the set, the light, and the framing exactly.\n- \n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "동일한 장소·의상·표정은 잘 지켰으나, 폰이 뒷면을 카메라로 향해 '화면이 인우 쪽과 부분적으로 카메라 쪽을 향하며 하단에서 메시지가 판독 가능'이라는 핵심 프레임 지시를 충족하지 못한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "얼굴·의상·슬링·배경을 그대로 유지한 채 폰 화면의 메시지 노출만 계약대로 교정했고, 시선·측면 여백·CCTV 모니터 연속성도 모두 충족해 얻은 것이 잃은 것보다 크다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.375,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.375,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "동일한 장소·의상·표정은 잘 지켰으나, 폰이 뒷면을 카메라로 향해 '화면이 인우 쪽과 부분적으로 카메라 쪽을 향하며 하단에서 메시지가 판독 가능'이라는 핵심 프레임 지시를 충족하지 못한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "얼굴·의상·슬링·배경을 그대로 유지한 채 폰 화면의 메시지 노출만 계약대로 교정했고, 시선·측면 여백·CCTV 모니터 연속성도 모두 충족해 얻은 것이 잃은 것보다 크다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "인우의 시선은 아래 오른쪽, 자신이 든 핸드폰 화면에 정확히 떨어진다. 화면은 위쪽으로 인우를 향하면서 동시에 카메라 쪽으로 살짝 기울어 메시지 말풍선(파란 대화창)이 판독 가능하게 보인다. 프레임 왼쪽에 인물이 향할 여백은 다소 적지만, 오른쪽 하단 프레임 경계에 폰이 놓여 샷 텍스트 요구를 충족. 배경 오른쪽에 CCTV 모니터 화면 일부가 초점 밖으로 남아 있다.",
      "built_space": "형사과 사무실: 왼쪽 서류더미와 캐비닛, 중앙 창문(새벽빛), 형광등 한 줄, 오른쪽 책상 위 보온병과 프린터·모니터. 이전 스틸의 벽면, 창틀, 천장 등기구, 책상 배치와 일관. 모니터는 하나만 있고 크기도 실제 스케일에 맞다.",
      "entities": "정인우: 한국인 남성, 짧은 검은 머리, 흰 셔츠에 남색 팔걸이 슬링 — 이전 스틸 및 캐릭터 레퍼런스와 일치(캐릭터 레퍼런스보다 나이가 조금 들어 보이나 이전 스틸과 동일 인물). 스마트폰: 검은 프레임 스마트폰, 화면이 위·카메라 쪽을 향하며 메시지 앱 UI 표시 — 프롭 레퍼런스와 부합. 배경 모니터는 CCTV 화면.",
      "hard_violations": [],
      "physics": "폰은 슬링을 하지 않은 오른손 손가락으로 자연스럽게 감싸 쥐고 있으며 엄지·검지의 그립이 보인다. 다친 팔은 슬링 안에 있고 몸통은 책상 쪽에 기대 앞으로 숙인 자세로 지지된다. 공중에 떠 있는 물체 없음."
     },
     {
      "label": "B",
      "direction": "시선은 아래로 폰을 향하지만, 카메라에 보이는 면은 폰의 뒷면(카메라 렌즈 모듈이 보이는 후면)이다. 즉 화면은 인우 쪽만 향하고 카메라 쪽으로는 전혀 열려 있지 않아 '메시지가 판독 가능해야 한다'는 지시가 실현되지 않는다. 배경 오른쪽 CCTV 모니터 일부 보임.",
      "built_space": "A와 동일한 사무실 세트(서류더미, 창문, 형광등, 보온병, 오른쪽 모니터). 구조적 오류 없음.",
      "entities": "정인우: 동일 인물, 흰 셔츠+남색 슬링으로 이전 스틸 일치. 스마트폰: 후면 렌즈 모듈이 보이는 검은 폰 — 프롭 레퍼런스의 외형과는 맞으나 기능면(화면)이 카메라에 전혀 보이지 않음. 표정은 A보다 완화되어 '얼굴이 잔뜩 일그러진' 강도가 약하다.",
      "hard_violations": [],
      "physics": "폰은 오른손에 쥐어져 있고 그립도 자연스럽다. 몸은 책상 쪽으로 숙인 채 지지됨. 부유 물체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "핸드폰 화면이 인우와 카메라 쪽을 동시에 향해 메시지가 하단 프레임에서 판독 가능하고, 미간과 입가가 잔뜩 일그러진 표정도 샷 텍스트에 부합 — 슬링·장소·저조도 조명 연속성도 유지."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "폰 후면만 카메라를 향해 '화면이 카메라 쪽으로도 향하며 메시지가 판독 가능해야 한다'는 프레임 지시를 놓쳤고, 표정 강도도 약해져 '도저히 이해할 수 없다는 듯 일그러진' 연기가 흐려졌다 — 배경·의상은 동일해 얻은 것 없이 핵심 항목만 잃었다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "핸드폰 화면이 인우와 카메라 쪽을 동시에 향해 메시지가 하단 프레임에서 판독 가능하고, 미간과 입가가 잔뜩 일그러진 표정도 샷 텍스트에 부합 — 슬링·장소·저조도 조명 연속성도 유지."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "폰 후면만 카메라를 향해 '화면이 카메라 쪽으로도 향하며 메시지가 판독 가능해야 한다'는 프레임 지시를 놓쳤고, 표정 강도도 약해져 '도저히 이해할 수 없다는 듯 일그러진' 연기가 흐려졌다 — 배경·의상은 동일해 얻은 것 없이 핵심 항목만 잃었다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "인우의 시선은 아래 오른쪽, 자신이 든 핸드폰 화면에 정확히 떨어진다. 화면은 위쪽으로 인우를 향하면서 동시에 카메라 쪽으로 살짝 기울어 메시지 말풍선(파란 대화창)이 판독 가능하게 보인다. 프레임 왼쪽에 인물이 향할 여백은 다소 적지만, 오른쪽 하단 프레임 경계에 폰이 놓여 샷 텍스트 요구를 충족. 배경 오른쪽에 CCTV 모니터 화면 일부가 초점 밖으로 남아 있다.",
      "built_space": "형사과 사무실: 왼쪽 서류더미와 캐비닛, 중앙 창문(새벽빛), 형광등 한 줄, 오른쪽 책상 위 보온병과 프린터·모니터. 이전 스틸의 벽면, 창틀, 천장 등기구, 책상 배치와 일관. 모니터는 하나만 있고 크기도 실제 스케일에 맞다.",
      "entities": "정인우: 한국인 남성, 짧은 검은 머리, 흰 셔츠에 남색 팔걸이 슬링 — 이전 스틸 및 캐릭터 레퍼런스와 일치(캐릭터 레퍼런스보다 나이가 조금 들어 보이나 이전 스틸과 동일 인물). 스마트폰: 검은 프레임 스마트폰, 화면이 위·카메라 쪽을 향하며 메시지 앱 UI 표시 — 프롭 레퍼런스와 부합. 배경 모니터는 CCTV 화면.",
      "hard_violations": [],
      "physics": "폰은 슬링을 하지 않은 오른손 손가락으로 자연스럽게 감싸 쥐고 있으며 엄지·검지의 그립이 보인다. 다친 팔은 슬링 안에 있고 몸통은 책상 쪽에 기대 앞으로 숙인 자세로 지지된다. 공중에 떠 있는 물체 없음."
     },
     {
      "label": "A",
      "direction": "시선은 아래로 폰을 향하지만, 카메라에 보이는 면은 폰의 뒷면(카메라 렌즈 모듈이 보이는 후면)이다. 즉 화면은 인우 쪽만 향하고 카메라 쪽으로는 전혀 열려 있지 않아 '메시지가 판독 가능해야 한다'는 지시가 실현되지 않는다. 배경 오른쪽 CCTV 모니터 일부 보임.",
      "built_space": "A와 동일한 사무실 세트(서류더미, 창문, 형광등, 보온병, 오른쪽 모니터). 구조적 오류 없음.",
      "entities": "정인우: 동일 인물, 흰 셔츠+남색 슬링으로 이전 스틸 일치. 스마트폰: 후면 렌즈 모듈이 보이는 검은 폰 — 프롭 레퍼런스의 외형과는 맞으나 기능면(화면)이 카메라에 전혀 보이지 않음. 표정은 A보다 완화되어 '얼굴이 잔뜩 일그러진' 강도가 약하다.",
      "hard_violations": [],
      "physics": "폰은 오른손에 쥐어져 있고 그립도 자연스럽다. 몸은 책상 쪽으로 숙인 채 지지됨. 부유 물체 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S16sh1"
  }
 },
 "S18sh3::bgfirst_bg": {
  "input_fingerprint": "279c960c23e17ce5",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 놀라 두 눈을 크게 뜬 채 운전석 쪽으로 고개를 돌려 고산들을 바라보는 김지후(강수리영)의 굳은 얼굴.\n\nLOCATION (lock): The front-seat cabin of the older SUV, between the passenger seat and driver's position.\n\nTIME OF DAY (lock): dawn.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At Jihoo's eye level near the space between the front seats, let the horizontal pan settle into a close three-quarter view exactly as her head finishes turning toward the driver. Jihoo's rigid face occupies most of the passenger-side frame, while a narrow, softly focused edge of Sandeul's driver-side shoulder preserves the conversational axis and her eyeline.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 구형 SUV 앞좌석 (두 인물이 각각 앞좌석에 앉아 있음) — The passenger-seat side opens toward Jihoo while the driver-seat edge remains close to the lens; used as The gap between the front seats creates a narrow visual corridor linking Jihoo to Sandeul.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Dawn ambient light holds Jihoo's widened eyes and clenched expression in subdued, high-contrast tones without introducing an additional visible source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 놀라 두 눈을 크게 뜬 채 운전석 쪽으로 고개를 돌려 고산들을 바라보는 김지후(강수리영)의 굳은 얼굴.\n\nLOCATION (lock): The front-seat cabin of the older SUV, between the passenger seat and driver's position.\n\nTIME OF DAY (lock): dawn.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At Jihoo's eye level near the space between the front seats, let the horizontal pan settle into a close three-quarter view exactly as her head finishes turning toward the driver. Jihoo's rigid face occupies most of the passenger-side frame, while a narrow, softly focused edge of Sandeul's driver-side shoulder preserves the conversational axis and her eyeline.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 구형 SUV 앞좌석 (두 인물이 각각 앞좌석에 앉아 있음) — The passenger-seat side opens toward Jihoo while the driver-seat edge remains close to the lens; used as The gap between the front seats creates a narrow visual corridor linking Jihoo to Sandeul.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Dawn ambient light holds Jihoo's widened eyes and clenched expression in subdued, high-contrast tones without introducing an additional visible source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S18sh3__bgfirst_bg.png",
  "asset_id": "a4c6abbe-2d65-4003-bd97-4919cfe1c03c",
  "input_asset_ids": [
   "56b10033-54e9-4c53-b191-3b2d1ca0401d",
   "29a0cf80-8d19-4faa-8450-e3db88ae02c4"
  ]
 },
 "S18sh3": {
  "input_fingerprint": "23ce328393cb1e08",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 놀라 두 눈을 크게 뜬 채 운전석 쪽으로 고개를 돌려 고산들을 바라보는 김지후(강수리영)의 굳은 얼굴.\n\nLOCATION (lock): The front-seat cabin of the older SUV, between the passenger seat and driver's position. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At Jihoo's eye level near the space between the front seats, let the horizontal pan settle into a close three-quarter view exactly as her head finishes turning toward the driver. Jihoo's rigid face occupies most of the passenger-side frame, while a narrow, softly focused edge of Sandeul's driver-side shoulder preserves the conversational axis and her eyeline.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 구형 SUV 앞좌석 (두 인물이 각각 앞좌석에 앉아 있음) — The passenger-seat side opens toward Jihoo while the driver-seat edge remains close to the lens; used as The gap between the front seats creates a narrow visual corridor linking Jihoo to Sandeul.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Dawn ambient light holds Jihoo's widened eyes and clenched expression in subdued, high-contrast tones without introducing an additional visible source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same pre-dawn SUV interior, framed photograph, and front-seat positions. Exclude a broad cabin view from the close framing; show her turning toward the driver in alarm.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu continues holding the same graduation photo frame taken from Minsuk’s home, and her necklace remains on.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot inside a tight, built interior. The FIRST attached image (SHOT BACKGROUND) is the finished empty interior of this shot, and in a space this cramped its geometry is the truth of the shot — keep it EXACTLY: its camera, perspective, every panel, control, seat, mirror, window and fixture stay untouched, in the same place, at the same angle, in the same number.\n\nBefore you place anyone, count what the background shows: how many steering wheels or control surfaces, how many seats and which way each faces, where each mirror sits and what it could reflect from this camera. Those counts and placements are what you must still be able to make after the people are in. Adding a second rim, sliding a seat, turning a mirror or growing a new panel is a failure even when the person looks right.\n\nThe SECOND attached image (LAYOUT SKETCH) tells you where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Its background lines are not decoration — they are the same structure seen in line form, so use them to place each person correctly with respect to it: which seat the body occupies, which side of the wheel the hands are on, what the body passes in front of and what it passes behind. Where sketch and background disagree about the structure itself, the background wins.\n\nThe CHARACTER REFERENCE photographs show the real people.\n\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. A person may cover part of the structure — that is expected, and covering is not redrawing. What the body hides stays hidden; what remains visible stays exactly as the background had it. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 놀라 두 눈을 크게 뜬 채 운전석 쪽으로 고개를 돌려 고산들을 바라보는 김지후(강수리영)의 굳은 얼굴.\n\nLOCATION (lock): The front-seat cabin of the older SUV, between the passenger seat and driver's position. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At Jihoo's eye level near the space between the front seats, let the horizontal pan settle into a close three-quarter view exactly as her head finishes turning toward the driver. Jihoo's rigid face occupies most of the passenger-side frame, while a narrow, softly focused edge of Sandeul's driver-side shoulder preserves the conversational axis and her eyeline.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 구형 SUV 앞좌석 (두 인물이 각각 앞좌석에 앉아 있음) — The passenger-seat side opens toward Jihoo while the driver-seat edge remains close to the lens; used as The gap between the front seats creates a narrow visual corridor linking Jihoo to Sandeul.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Dawn ambient light holds Jihoo's widened eyes and clenched expression in subdued, high-contrast tones without introducing an additional visible source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu continues holding the same graduation photo frame taken from Minsuk’s home, and her necklace remains on.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): dawn.\n\nSHOT TEXT (authoritative, Korean): 놀라 두 눈을 크게 뜬 채 운전석 쪽으로 고개를 돌려 고산들을 바라보는 김지후(강수리영)의 굳은 얼굴.\n\nLOCATION (lock): The front-seat cabin of the older SUV, between the passenger seat and driver's position. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At Jihoo's eye level near the space between the front seats, let the horizontal pan settle into a close three-quarter view exactly as her head finishes turning toward the driver. Jihoo's rigid face occupies most of the passenger-side frame, while a narrow, softly focused edge of Sandeul's driver-side shoulder preserves the conversational axis and her eyeline.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 구형 SUV 앞좌석 (두 인물이 각각 앞좌석에 앉아 있음) — The passenger-seat side opens toward Jihoo while the driver-seat edge remains close to the lens; used as The gap between the front seats creates a narrow visual corridor linking Jihoo to Sandeul.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Dawn ambient light holds Jihoo's widened eyes and clenched expression in subdued, high-contrast tones without introducing an additional visible source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same pre-dawn SUV interior, framed photograph, and front-seat positions. Exclude a broad cabin view from the close framing; show her turning toward the driver in alarm.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu continues holding the same graduation photo frame taken from Minsuk’s home, and her necklace remains on.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S18sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement AND the structure they sit inside — the sketched panels, seats, controls and openings are the same ones the background photograph shows, drawn as lines; read them to place each body correctly against that structure, and where the two disagree about the structure itself the background photograph wins)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S18sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:904317>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L12B01.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:904317>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "지후는 조수석에 앉아 고개를 왼쪽(운전석 방향)으로 돌리고 있고, 시선은 프레임 왼쪽 전경의 남성 뒤통수/어깨 쪽으로 정확히 향한다. 눈은 크게 뜬 채 놀란 표정. 대화 축(지후→운전석)이 좌석 사이 통로를 따라 성립한다.",
      "built_space": "구형 SUV 앞좌석 실내. 스티어링 휠은 좌측에 하나만 보이고 계기판, 센터페시아 오디오(디지털 시계), 에어컨 송풍구, 오토 기어 셀렉터, 조수석 도어 트림과 사이드미러가 레퍼런스 실내와 일치한다. 남성은 좌측 운전석, 지후는 우측 조수석 — 좌석 배치가 스테이징과 맞다. 운전석 어깨는 전경에서 흐릿하게 걸려 있다.",
      "entities": "20대 초반 한국 여성, 긴 검은 생머리, 회색 티셔츠 위 파란 안전조끼 — 캐릭터 레퍼런스와 일치. 목에는 가는 은목걸이. 손에는 졸업사진 액자를 뒷면(합판+걸이 금구)이 카메라 쪽으로 오게 안고 있어 기능면은 본인 쪽 — 프롭 방향 규칙에 부합. 남성은 뒤통수/어깨만 보이는 부분 노출.",
      "hard_violations": [],
      "physics": "지후는 조수석 시트에 앉아 있고 등과 엉덩이가 시트에 접촉. 액자는 오른손이 하단 모서리를 감싸고 가슴/무릎 사이에 받쳐져 있어 지지점이 명확. 떠 있는 물체 없음. 창밖 도로면과 차체 원근이 자연스럽다."
     },
     {
      "label": "B",
      "direction": "지후의 시선은 프레임 오른쪽 남성 쪽으로 향하고 고개도 그쪽으로 돌아가 있어 아이라인은 성립. 다만 지후가 앉은 자리는 앞좌석이 아니라 두 앞좌석 헤드레스트 뒤쪽(뒷좌석/2열)이며, 카메라도 뒷좌석 축에서 잡는다.",
      "built_space": "차량 실내이나 앞좌석 등받이 두 개가 프레임 좌·중앙에 뒤에서 보이고, 지후는 그 뒤 공간에 있다. 스티어링 휠·대시보드·계기판·오디오 등 레퍼런스의 고정 특징이 전혀 보이지 않아 '앞좌석 캐빈, 조수석과 운전석 사이' 로케이션 락과 불일치. 남성은 우측 앞좌석 등받이 옆에 있어 좌/우 배치도 반전. 창밖은 새벽이라기보다 밝은 흐린 낮에 가깝다.",
      "entities": "20대 초반 한국 여성, 긴 검은 머리, 회색 티셔츠와 파란 조끼 어깨끈, 은목걸이 펜던트 확인 — 인물 일치. 액자는 뒷면이 카메라를 향해 손에 들려 있다. 남성은 뒤통수/어깨만.",
      "hard_violations": [
       "지후가 앞좌석(조수석)이 아닌 뒷좌석 위치에 앉아 있어 스테이징이 지정하지 않은 자리에 배치됨 — 로케이션/좌석 배치 위반"
      ],
      "physics": "지후는 좌석에 앉아 있고 액자는 왼손이 하단을 잡아 지지. 공중에 뜬 요소는 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "조수석의 지후가 운전석 쪽으로 고개를 돌려 놀란 눈으로 바라보고, 레퍼런스 SUV 앞좌석 실내(휠·대시·기어)와 좌석 배치, 액자·목걸이가 모두 일치 — 다소 여유 있는 미디엄 클로즈업이지만 지정 구도와 로케이션을 정확히 실현."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "얼굴 클로즈업과 표정은 좋으나 지후가 앞좌석이 아닌 뒷좌석에 앉아 있고 대시보드·핸들 등 로케이션 고정 특징이 전혀 없으며 광량도 새벽보다 낮에 가까워 로케이션 락 위반."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "조수석의 지후가 운전석 쪽으로 고개를 돌려 놀란 눈으로 바라보고, 레퍼런스 SUV 앞좌석 실내(휠·대시·기어)와 좌석 배치, 액자·목걸이가 모두 일치 — 다소 여유 있는 미디엄 클로즈업이지만 지정 구도와 로케이션을 정확히 실현."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "얼굴 클로즈업과 표정은 좋으나 지후가 앞좌석이 아닌 뒷좌석에 앉아 있고 대시보드·핸들 등 로케이션 고정 특징이 전혀 없으며 광량도 새벽보다 낮에 가까워 로케이션 락 위반."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "지후는 조수석에 앉아 고개를 왼쪽(운전석 방향)으로 돌리고 있고, 시선은 프레임 왼쪽 전경의 남성 뒤통수/어깨 쪽으로 정확히 향한다. 눈은 크게 뜬 채 놀란 표정. 대화 축(지후→운전석)이 좌석 사이 통로를 따라 성립한다.",
      "built_space": "구형 SUV 앞좌석 실내. 스티어링 휠은 좌측에 하나만 보이고 계기판, 센터페시아 오디오(디지털 시계), 에어컨 송풍구, 오토 기어 셀렉터, 조수석 도어 트림과 사이드미러가 레퍼런스 실내와 일치한다. 남성은 좌측 운전석, 지후는 우측 조수석 — 좌석 배치가 스테이징과 맞다. 운전석 어깨는 전경에서 흐릿하게 걸려 있다.",
      "entities": "20대 초반 한국 여성, 긴 검은 생머리, 회색 티셔츠 위 파란 안전조끼 — 캐릭터 레퍼런스와 일치. 목에는 가는 은목걸이. 손에는 졸업사진 액자를 뒷면(합판+걸이 금구)이 카메라 쪽으로 오게 안고 있어 기능면은 본인 쪽 — 프롭 방향 규칙에 부합. 남성은 뒤통수/어깨만 보이는 부분 노출.",
      "hard_violations": [],
      "physics": "지후는 조수석 시트에 앉아 있고 등과 엉덩이가 시트에 접촉. 액자는 오른손이 하단 모서리를 감싸고 가슴/무릎 사이에 받쳐져 있어 지지점이 명확. 떠 있는 물체 없음. 창밖 도로면과 차체 원근이 자연스럽다."
     },
     {
      "label": "B",
      "direction": "지후의 시선은 프레임 오른쪽 남성 쪽으로 향하고 고개도 그쪽으로 돌아가 있어 아이라인은 성립. 다만 지후가 앉은 자리는 앞좌석이 아니라 두 앞좌석 헤드레스트 뒤쪽(뒷좌석/2열)이며, 카메라도 뒷좌석 축에서 잡는다.",
      "built_space": "차량 실내이나 앞좌석 등받이 두 개가 프레임 좌·중앙에 뒤에서 보이고, 지후는 그 뒤 공간에 있다. 스티어링 휠·대시보드·계기판·오디오 등 레퍼런스의 고정 특징이 전혀 보이지 않아 '앞좌석 캐빈, 조수석과 운전석 사이' 로케이션 락과 불일치. 남성은 우측 앞좌석 등받이 옆에 있어 좌/우 배치도 반전. 창밖은 새벽이라기보다 밝은 흐린 낮에 가깝다.",
      "entities": "20대 초반 한국 여성, 긴 검은 머리, 회색 티셔츠와 파란 조끼 어깨끈, 은목걸이 펜던트 확인 — 인물 일치. 액자는 뒷면이 카메라를 향해 손에 들려 있다. 남성은 뒤통수/어깨만.",
      "hard_violations": [
       "지후가 앞좌석(조수석)이 아닌 뒷좌석 위치에 앉아 있어 스테이징이 지정하지 않은 자리에 배치됨 — 로케이션/좌석 배치 위반"
      ],
      "physics": "지후는 좌석에 앉아 있고 액자는 왼손이 하단을 잡아 지지. 공중에 뜬 요소는 없음."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "지후의 시선과 얼굴은 프레임 오른쪽 전경의 남성(고산들) 뒤통수·어깨 쪽으로 정확히 향해 있다. 고개는 운전석 방향(화면 오른쪽)으로 돌아가 있고 눈은 크게 뜬 채 그 인물에 고정. 눈맞춤 축이 성립한다.",
      "built_space": "차량 실내이지만 앞좌석 사이 통로가 아니라 뒷좌석 쪽에서 본 듯한 배치로 읽힌다. 전경 왼쪽에 앞좌석 헤드레스트, 오른쪽에 운전석 등받이와 남성 뒤통수가 있으나 대시보드·스티어링휠은 보이지 않고 지후는 뒷좌석에 앉은 듯 보인다. 스티어링휠은 프레임 밖(중복 없음). 좌석 수·배치 자체에 물리적 모순은 없다.",
      "entities": "지후: 20대 초반 한국 여성, 긴 검은 생머리, 캐릭터 레퍼런스와 유사한 얼굴. 파란 조끼 어깨 일부와 회색 티셔츠, 목걸이(펜던트) 확인됨. 손에 든 액자는 뒷면(합판·걸이 철물)이 카메라를 향해 있어 기능면이 본인 쪽 — 소품 방향 규칙에 부합. 운전석 남성은 뒤통수·어깨만 보여 익명 처리, 대사 축 유지용.",
      "hard_violations": [],
      "physics": "지후는 좌석에 앉아 있고 액자는 오른손이 아래 모서리를 확실히 쥐고 몸에 기대어 지지된다. 남성은 좌석에 앉은 자세. 공중에 뜬 물체나 지지 없는 신체 없음."
     },
     {
      "label": "B",
      "direction": "지후의 얼굴은 화면 왼쪽 전방(앞유리/운전석 방향)을 향하지만 시선이 왼쪽 전경 남성의 머리보다 위·앞쪽으로 지나가 인물에 정확히 안착하지 않는 느낌. 고개는 운전석 쪽으로 돌아 있어 방향성 자체는 맞다.",
      "built_space": "구형 SUV 앞좌석 실내가 레퍼런스와 매우 잘 일치(대시보드, 오디오, 시프트 레버, 사이드미러, 도로 배경). 스티어링휠 1개, 운전석에 남성 1명, 조수석에 지후. 그러나 카메라가 운전석 뒤/어깨 너머 광각으로 물러나 있어 '조수석-운전석 사이 눈높이 클로즈업'이 아니라 미디엄~와이드 캐빈 전경이 되었다. 이는 '넓은 캐빈 뷰 제외' 지시 위반.",
      "entities": "지후: 한국 여성, 긴 검은 머리, 파란 안전조끼+회색 티셔츠, 목걸이 확인. 다만 얼굴이 레퍼런스보다 다소 나이 들어 보이고 얼굴 비율이 약간 다름. 액자는 뒷면이 카메라 쪽 — 방향 규칙 부합. 배경은 레퍼런스 도로 풍경과 일치. 오디오 시계에 '15:50' 숫자가 보이며 하늘은 여명이라 시간 표시가 어긋난다.",
      "hard_violations": [],
      "physics": "지후는 조수석에 앉아 액자를 두 손/오른손으로 감싸 쥐고 무릎·몸통에 기대 지지. 남성은 운전석에 앉아 핸들 쪽에 있음. 부유하는 물체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "고개를 운전석 쪽으로 돌린 지후의 굳은 얼굴이 프레임 대부분을 차지하는 클로즈 3/4 구도이며 운전석 어깨가 흐릿하게 대화 축을 지켜 카메라 지시에 가장 부합; 좌석 위치가 다소 뒤쪽으로 읽히고 대시보드 등 로케이션 단서가 약한 점이 감점."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "구형 SUV 실내 로케이션 재현은 우수하나 '넓은 캐빈 뷰 제외·클로즈업' 지시를 어긴 미디엄 와이드이며 인물이 작고 시선이 상대에게 정확히 안착하지 않고 시계 표시(15:50)가 여명과 충돌."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "고개를 운전석 쪽으로 돌린 지후의 굳은 얼굴이 프레임 대부분을 차지하는 클로즈 3/4 구도이며 운전석 어깨가 흐릿하게 대화 축을 지켜 카메라 지시에 가장 부합; 좌석 위치가 다소 뒤쪽으로 읽히고 대시보드 등 로케이션 단서가 약한 점이 감점."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "구형 SUV 실내 로케이션 재현은 우수하나 '넓은 캐빈 뷰 제외·클로즈업' 지시를 어긴 미디엄 와이드이며 인물이 작고 시선이 상대에게 정확히 안착하지 않고 시계 표시(15:50)가 여명과 충돌."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "지후의 시선과 얼굴은 프레임 오른쪽 전경의 남성(고산들) 뒤통수·어깨 쪽으로 정확히 향해 있다. 고개는 운전석 방향(화면 오른쪽)으로 돌아가 있고 눈은 크게 뜬 채 그 인물에 고정. 눈맞춤 축이 성립한다.",
      "built_space": "차량 실내이지만 앞좌석 사이 통로가 아니라 뒷좌석 쪽에서 본 듯한 배치로 읽힌다. 전경 왼쪽에 앞좌석 헤드레스트, 오른쪽에 운전석 등받이와 남성 뒤통수가 있으나 대시보드·스티어링휠은 보이지 않고 지후는 뒷좌석에 앉은 듯 보인다. 스티어링휠은 프레임 밖(중복 없음). 좌석 수·배치 자체에 물리적 모순은 없다.",
      "entities": "지후: 20대 초반 한국 여성, 긴 검은 생머리, 캐릭터 레퍼런스와 유사한 얼굴. 파란 조끼 어깨 일부와 회색 티셔츠, 목걸이(펜던트) 확인됨. 손에 든 액자는 뒷면(합판·걸이 철물)이 카메라를 향해 있어 기능면이 본인 쪽 — 소품 방향 규칙에 부합. 운전석 남성은 뒤통수·어깨만 보여 익명 처리, 대사 축 유지용.",
      "hard_violations": [],
      "physics": "지후는 좌석에 앉아 있고 액자는 오른손이 아래 모서리를 확실히 쥐고 몸에 기대어 지지된다. 남성은 좌석에 앉은 자세. 공중에 뜬 물체나 지지 없는 신체 없음."
     },
     {
      "label": "A",
      "direction": "지후의 얼굴은 화면 왼쪽 전방(앞유리/운전석 방향)을 향하지만 시선이 왼쪽 전경 남성의 머리보다 위·앞쪽으로 지나가 인물에 정확히 안착하지 않는 느낌. 고개는 운전석 쪽으로 돌아 있어 방향성 자체는 맞다.",
      "built_space": "구형 SUV 앞좌석 실내가 레퍼런스와 매우 잘 일치(대시보드, 오디오, 시프트 레버, 사이드미러, 도로 배경). 스티어링휠 1개, 운전석에 남성 1명, 조수석에 지후. 그러나 카메라가 운전석 뒤/어깨 너머 광각으로 물러나 있어 '조수석-운전석 사이 눈높이 클로즈업'이 아니라 미디엄~와이드 캐빈 전경이 되었다. 이는 '넓은 캐빈 뷰 제외' 지시 위반.",
      "entities": "지후: 한국 여성, 긴 검은 머리, 파란 안전조끼+회색 티셔츠, 목걸이 확인. 다만 얼굴이 레퍼런스보다 다소 나이 들어 보이고 얼굴 비율이 약간 다름. 액자는 뒷면이 카메라 쪽 — 방향 규칙 부합. 배경은 레퍼런스 도로 풍경과 일치. 오디오 시계에 '15:50' 숫자가 보이며 하늘은 여명이라 시간 표시가 어긋난다.",
      "hard_violations": [],
      "physics": "지후는 조수석에 앉아 액자를 두 손/오른손으로 감싸 쥐고 무릎·몸통에 기대 지지. 남성은 운전석에 앉아 핸들 쪽에 있음. 부유하는 물체 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 13,
     "B": 12
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "readings": [
   {
    "label": "A",
    "direction": "지후는 조수석에 앉아 고개를 왼쪽(운전석 방향)으로 돌리고 있고, 시선은 프레임 왼쪽 전경의 남성 뒤통수/어깨 쪽으로 정확히 향한다. 눈은 크게 뜬 채 놀란 표정. 대화 축(지후→운전석)이 좌석 사이 통로를 따라 성립한다.",
    "built_space": "구형 SUV 앞좌석 실내. 스티어링 휠은 좌측에 하나만 보이고 계기판, 센터페시아 오디오(디지털 시계), 에어컨 송풍구, 오토 기어 셀렉터, 조수석 도어 트림과 사이드미러가 레퍼런스 실내와 일치한다. 남성은 좌측 운전석, 지후는 우측 조수석 — 좌석 배치가 스테이징과 맞다. 운전석 어깨는 전경에서 흐릿하게 걸려 있다.",
    "entities": "20대 초반 한국 여성, 긴 검은 생머리, 회색 티셔츠 위 파란 안전조끼 — 캐릭터 레퍼런스와 일치. 목에는 가는 은목걸이. 손에는 졸업사진 액자를 뒷면(합판+걸이 금구)이 카메라 쪽으로 오게 안고 있어 기능면은 본인 쪽 — 프롭 방향 규칙에 부합. 남성은 뒤통수/어깨만 보이는 부분 노출.",
    "hard_violations": [],
    "physics": "지후는 조수석 시트에 앉아 있고 등과 엉덩이가 시트에 접촉. 액자는 오른손이 하단 모서리를 감싸고 가슴/무릎 사이에 받쳐져 있어 지지점이 명확. 떠 있는 물체 없음. 창밖 도로면과 차체 원근이 자연스럽다."
   },
   {
    "label": "B",
    "direction": "지후의 시선은 프레임 오른쪽 남성 쪽으로 향하고 고개도 그쪽으로 돌아가 있어 아이라인은 성립. 다만 지후가 앉은 자리는 앞좌석이 아니라 두 앞좌석 헤드레스트 뒤쪽(뒷좌석/2열)이며, 카메라도 뒷좌석 축에서 잡는다.",
    "built_space": "차량 실내이나 앞좌석 등받이 두 개가 프레임 좌·중앙에 뒤에서 보이고, 지후는 그 뒤 공간에 있다. 스티어링 휠·대시보드·계기판·오디오 등 레퍼런스의 고정 특징이 전혀 보이지 않아 '앞좌석 캐빈, 조수석과 운전석 사이' 로케이션 락과 불일치. 남성은 우측 앞좌석 등받이 옆에 있어 좌/우 배치도 반전. 창밖은 새벽이라기보다 밝은 흐린 낮에 가깝다.",
    "entities": "20대 초반 한국 여성, 긴 검은 머리, 회색 티셔츠와 파란 조끼 어깨끈, 은목걸이 펜던트 확인 — 인물 일치. 액자는 뒷면이 카메라를 향해 손에 들려 있다. 남성은 뒤통수/어깨만.",
    "hard_violations": [
     "지후가 앞좌석(조수석)이 아닌 뒷좌석 위치에 앉아 있어 스테이징이 지정하지 않은 자리에 배치됨 — 로케이션/좌석 배치 위반"
    ],
    "physics": "지후는 좌석에 앉아 있고 액자는 왼손이 하단을 잡아 지지. 공중에 뜬 요소는 없음."
   }
  ],
  "totals": {
   "A": 13,
   "B": 12
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "조수석의 지후가 운전석 쪽으로 고개를 돌려 놀란 눈으로 바라보고, 레퍼런스 SUV 앞좌석 실내(휠·대시·기어)와 좌석 배치, 액자·목걸이가 모두 일치 — 다소 여유 있는 미디엄 클로즈업이지만 지정 구도와 로케이션을 정확히 실현."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "얼굴 클로즈업과 표정은 좋으나 지후가 앞좌석이 아닌 뒷좌석에 앉아 있고 대시보드·핸들 등 로케이션 고정 특징이 전혀 없으며 광량도 새벽보다 낮에 가까워 로케이션 락 위반."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L12B01.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:904317>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "원본 배경 및 레이아웃 스케치에 없는 룸미러가 유리창 상단 중앙에 임의로 생성되었습니다.",
     "fix_en": "Remove the rear-view mirror from the top center of the windshield, replacing it with the continuous dawn sky and distant city skyline, while strictly preserving both characters, their poses, clothing, the picture frame, and all other car interior structures exactly as they are.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the rear-view mirror from the top center of the windshield, replacing it with the continuous dawn sky and distant city skyline, while strictly preserving both characters, their poses, clothing, the picture frame, and all other car interior structures exactly as they are.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "시선·좌석 배치·액자 방향은 맞지만, 배경 사진에 없는 실내 룸미러를 상단에 새로 만들고 좌측 헤드레스트 형태를 바꿔 '배경 기하 그대로 유지' 락을 위반했으며, 얼굴도 레퍼런스보다 성숙해져 계약이 요구하지 않은 손실이 발생했다.  ★위반: [claude-opus] 배경 사진에 없는 실내 룸미러가 프레임 상단에 새로 생성됨(고정물 추가) / [gpt] 기준 배경에 보이지 않던 룸미러를 화면 상단 중앙에 새로 추가해, 고정된 차량 내부의 거울 개수와 배치를 변경했다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "배경의 스티어링 휠·계기판·센터페시아·미러 수와 위치를 하나도 늘리거나 옮기지 않은 채 스케치대로 조수석 지후가 운전석 쪽으로 고개를 돌린 클로즈업을 구현했고, 액자·목걸이·의상·새벽광까지 유지해 변경량이 가장 적다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.25,
      "B": 2.0
     },
     "adjusted": {
      "A": 0.75,
      "B": 2.0
     },
     "violations": {
      "A": [
       "[claude-opus] 배경 사진에 없는 실내 룸미러가 프레임 상단에 새로 생성됨(고정물 추가)",
       "[gpt] 기준 배경에 보이지 않던 룸미러를 화면 상단 중앙에 새로 추가해, 고정된 차량 내부의 거울 개수와 배치를 변경했다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "시선·좌석 배치·액자 방향은 맞지만, 배경 사진에 없는 실내 룸미러를 상단에 새로 만들고 좌측 헤드레스트 형태를 바꿔 '배경 기하 그대로 유지' 락을 위반했으며, 얼굴도 레퍼런스보다 성숙해져 계약이 요구하지 않은 손실이 발생했다.  ★위반: [claude-opus] 배경 사진에 없는 실내 룸미러가 프레임 상단에 새로 생성됨(고정물 추가) / [gpt] 기준 배경에 보이지 않던 룸미러를 화면 상단 중앙에 새로 추가해, 고정된 차량 내부의 거울 개수와 배치를 변경했다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "배경의 스티어링 휠·계기판·센터페시아·미러 수와 위치를 하나도 늘리거나 옮기지 않은 채 스케치대로 조수석 지후가 운전석 쪽으로 고개를 돌린 클로즈업을 구현했고, 액자·목걸이·의상·새벽광까지 유지해 변경량이 가장 적다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "readings": [
     {
      "label": "A",
      "direction": "지후는 조수석에 앉아 상체와 고개를 왼쪽(운전석 쪽)으로 돌리고 있으며, 시선이 프레임 좌측 전경의 남자(고산들) 뒤통수·어깨 방향으로 정확히 향한다. 운전석 인물은 정면(도로) 쪽을 보고 있어 대화 축이 성립. 화살표 방향(지후 얼굴→좌측 운전석)과 일치.",
      "built_space": "스티어링 휠 1개(좌측, 운전석 앞)만 존재. 계기판·오디오(시계 표시)·센터 콘솔·기어 셀렉터·글로브박스·조수석 도어 트림·사이드미러 1개 모두 배경 사진과 같은 위치·각도. 조수석에 지후, 운전석에 산들 각 1명씩. 후방 조수석 헤드레스트가 우하단에 배경대로 남아 있음. 실내등/룸미러 추가 없음.",
      "entities": "지후: 한국인 20대 초반 여성, 긴 검은 생머리, 캐릭터 레퍼런스의 파란 형광 조끼+회색 티셔츠 일치. 목걸이 가시적(계약의 necklace 유지). 졸업사진 액자를 뒷면(갈색 백보드, 걸이 고리)이 카메라를 향하게 두 손으로 안고 있어 '기능면은 본인 쪽' 규칙과 부합. 좌측 남성은 뒤통수·어깨만 부드럽게 보이는 좁은 가장자리로 처리됨.",
      "hard_violations": [],
      "physics": "지후는 조수석 시트에 앉아 등받이에 기대고, 액자는 오른손이 아래 모서리를 잡고 왼팔로 가슴에 받쳐 안겨 있어 지지가 명확. 남자는 운전석 시트에 앉아 헤드레스트/시트에 몸이 걸쳐 있음. 공중에 뜬 물체 없음."
     },
     {
      "label": "B",
      "direction": "지후의 고개·시선은 좌측 운전석 남자 쪽을 향하나 A보다 약간 정면/전방으로 흐르는 인상. 남자는 좌측에서 정면을 향함. 아이라인은 대체로 성립.",
      "built_space": "스티어링 휠 1개. 그러나 프레임 상단 중앙에 배경 사진에는 없는 룸미러(백미러)가 새로 생겨 상단을 크게 침범하고 있으며, 프레임이 전체적으로 아래로 내려가/좁아져 윈드실드 상단 프레임 구조가 달라졌다. 조수석 도어·사이드미러·오디오·기어는 배경대로. 앞유리 상단에 없던 미러 마운트 패널이 추가된 것은 배경 구조 개변.",
      "entities": "지후: 동일 인물, 긴 검은 머리, 파란 조끼+회색 티, 목걸이 있음. 액자는 뒷면이 카메라를 향한 상태로 A와 동일. 좌측 남성은 뒤통수·어깨만.",
      "hard_violations": [
       "배경 사진에 존재하지 않는 룸미러(백미러)와 마운트가 프레임 상단에 새로 생성됨 — 고정된 구조물 개수/배치 변경"
      ],
      "physics": "지후는 시트에 앉아 오른손으로 액자 하단을 받쳐 안고 있으며 지지 명확. 남자도 운전석에 착석. 공중 부양 요소 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "배경 SUV 실내 구조(휠 1개, 미러·계기·콘솔 위치)를 그대로 유지한 채 지후가 조수석에서 운전석 쪽으로 고개를 돌려 놀란 표정을 짓는 순간을 스케치 배치대로 실현했고, 액자 뒷면·목걸이·조끼 등 지속 소품도 충족한다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "인물·소품·시선은 A와 거의 같지만, 배경에 없던 룸미러와 상단 마운트를 새로 만들어 넣고 프레임을 아래로 밀어 실내 구조 고정 조건을 어겼으며, 표정의 '두 눈을 크게 뜬' 강도도 다소 약해 얻은 것보다 잃은 것이 크다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "배경 SUV 실내 구조(휠 1개, 미러·계기·콘솔 위치)를 그대로 유지한 채 지후가 조수석에서 운전석 쪽으로 고개를 돌려 놀란 표정을 짓는 순간을 스케치 배치대로 실현했고, 액자 뒷면·목걸이·조끼 등 지속 소품도 충족한다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "인물·소품·시선은 A와 거의 같지만, 배경에 없던 룸미러와 상단 마운트를 새로 만들어 넣고 프레임을 아래로 밀어 실내 구조 고정 조건을 어겼으며, 표정의 '두 눈을 크게 뜬' 강도도 다소 약해 얻은 것보다 잃은 것이 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "지후는 조수석에 앉아 상체와 고개를 왼쪽(운전석 쪽)으로 돌리고 있으며, 시선이 프레임 좌측 전경의 남자(고산들) 뒤통수·어깨 방향으로 정확히 향한다. 운전석 인물은 정면(도로) 쪽을 보고 있어 대화 축이 성립. 화살표 방향(지후 얼굴→좌측 운전석)과 일치.",
      "built_space": "스티어링 휠 1개(좌측, 운전석 앞)만 존재. 계기판·오디오(시계 표시)·센터 콘솔·기어 셀렉터·글로브박스·조수석 도어 트림·사이드미러 1개 모두 배경 사진과 같은 위치·각도. 조수석에 지후, 운전석에 산들 각 1명씩. 후방 조수석 헤드레스트가 우하단에 배경대로 남아 있음. 실내등/룸미러 추가 없음.",
      "entities": "지후: 한국인 20대 초반 여성, 긴 검은 생머리, 캐릭터 레퍼런스의 파란 형광 조끼+회색 티셔츠 일치. 목걸이 가시적(계약의 necklace 유지). 졸업사진 액자를 뒷면(갈색 백보드, 걸이 고리)이 카메라를 향하게 두 손으로 안고 있어 '기능면은 본인 쪽' 규칙과 부합. 좌측 남성은 뒤통수·어깨만 부드럽게 보이는 좁은 가장자리로 처리됨.",
      "hard_violations": [],
      "physics": "지후는 조수석 시트에 앉아 등받이에 기대고, 액자는 오른손이 아래 모서리를 잡고 왼팔로 가슴에 받쳐 안겨 있어 지지가 명확. 남자는 운전석 시트에 앉아 헤드레스트/시트에 몸이 걸쳐 있음. 공중에 뜬 물체 없음."
     },
     {
      "label": "A",
      "direction": "지후의 고개·시선은 좌측 운전석 남자 쪽을 향하나 A보다 약간 정면/전방으로 흐르는 인상. 남자는 좌측에서 정면을 향함. 아이라인은 대체로 성립.",
      "built_space": "스티어링 휠 1개. 그러나 프레임 상단 중앙에 배경 사진에는 없는 룸미러(백미러)가 새로 생겨 상단을 크게 침범하고 있으며, 프레임이 전체적으로 아래로 내려가/좁아져 윈드실드 상단 프레임 구조가 달라졌다. 조수석 도어·사이드미러·오디오·기어는 배경대로. 앞유리 상단에 없던 미러 마운트 패널이 추가된 것은 배경 구조 개변.",
      "entities": "지후: 동일 인물, 긴 검은 머리, 파란 조끼+회색 티, 목걸이 있음. 액자는 뒷면이 카메라를 향한 상태로 A와 동일. 좌측 남성은 뒤통수·어깨만.",
      "hard_violations": [
       "배경 사진에 존재하지 않는 룸미러(백미러)와 마운트가 프레임 상단에 새로 생성됨 — 고정된 구조물 개수/배치 변경"
      ],
      "physics": "지후는 시트에 앉아 오른손으로 액자 하단을 받쳐 안고 있으며 지지 명확. 남자도 운전석에 착석. 공중 부양 요소 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S18sh3__bgfirst_bg.png",
   "bg_asset_id": "a4c6abbe-2d65-4003-bd97-4919cfe1c03c",
   "bg_record_key": "S18sh3::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S18sh1"
  }
 },
 "S20sh13": {
  "input_fingerprint": "4a6353bd2551b9df",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)의 시선을 외면한 채 마을 쪽으로 등을 돌려 걷는 mid-stride 자세로 한쪽 발이 땅에서 떨어져 있는 고산들의 뒷모습.\n\nLOCATION (lock): The landward walkway leading from the island pier into the coastal village. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track from adult shoulder height several meters behind Sandeul, holding a wide view that keeps his complete back-facing body in the middle of the frame as one lifted foot marks the stride toward the village. Jihoo remains confined to a soft lower-left foreground edge, watching him leave, while the pier's route draws forward toward the settlement.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 고산들 in the middle-center of the frame, midground, moves toward village; village in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 금월도 선착장 (마을 쪽으로 이어져 있음) — Its long axis points from the camera and Jihoo toward the village ahead of Sandeul; used as The pier forms the directional path of Sandeul's retreat; 작은 섬 마을의 집들 (알록달록한 지붕의 집들이 모여 있음) — Their colorful rooflines face the approaching path from the pier; used as The settlement serves as Sandeul's destination in the upper background.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime light maintains the welcoming village surface while restrained contrast turns Sandeul's retreating silhouette emotionally cold.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same pier, village approach, bright daylight, and continuing welcoming crowd. Exclude the embrace and center the man walking away toward the village.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu retains the white animal-bone necklace as Sandel walks away from her.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (Korean man in his mid-40s, closely cropped dark hair, cold angular face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 9,
   "C": 7,
   "B": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 9,
    "verdict_ko": "등을 보인 고산들이 화면 중앙 중경에서 한 발을 뗀 mid-stride로 마을 쪽으로 걸어가고, 지후는 좌하단 전경에서 흰 뼈 목걸이를 유지한 채 그를 바라보며, 부두 장축·태극기 환영 인파·알록달록한 마을 지붕이 이전 스틸과 정확히 이어진다."
   },
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "부두 방향성과 mid-stride 지지는 맞으나, 지후의 시선이 그를 향하지 않고 배경이 산·나무 위주라 '상단 중앙의 알록달록한 마을'과 환영 인파의 연속성이 약하다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "인물 의상이 캐릭터 레퍼런스의 올리브 조끼+체크 셔츠가 아닌 단색 재킷으로 어긋나고, 인파가 길을 막아 마을로 향하는 퇴장 경로가 흐려지며 마을 지붕이 상단 중앙에서 밀려났다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S20sh11_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903959>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "카메라를 등지고 있는 지후의 등에 뼈 목걸이가 잘못 그려져 있습니다. 목걸이 펜던트는 몸 앞쪽에 있어야 하므로 뒷모습에서는 보이지 않아야 합니다.",
     "fix_en": "Remove the white bone necklace and its strings from the back of the woman in the left foreground, restoring the continuous fabric pattern of her shirt in that space. Preserve the woman's silhouette and soft focus, the walking man, his pose and clothing, the background crowd, the village, the lighting, and the exact camera framing.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the white bone necklace and its strings from the back of the woman in the left foreground, restoring the continuous fabric pattern of her shirt in that space. Preserve the woman's silhouette and soft focus, the walking man, his pose and clothing, the background crowd, the village, the lighting, and the exact camera framing.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "뒤에서 잡은 와이드, 중앙의 산들 mid-stride, 좌하단 전경 지후, 마을로 향하는 선착장 축이 모두 충족되고 지후의 흰 동물뼈 목걸이(CARRIED STATE)까지 명확히 보존 — 계약 항목을 가장 많이 지킨다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "구도·자세·장소는 A와 사실상 동일하나 지후의 흰 동물뼈 목걸이가 사라졌고 전경 인물이 커져 '소프트한 좌하단 가장자리' 규정에서 멀어짐 — 계약이 요구하지 않은 변경으로 손실만 발생."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.75
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.75
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "뒤에서 잡은 와이드, 중앙의 산들 mid-stride, 좌하단 전경 지후, 마을로 향하는 선착장 축이 모두 충족되고 지후의 흰 동물뼈 목걸이(CARRIED STATE)까지 명확히 보존 — 계약 항목을 가장 많이 지킨다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "구도·자세·장소는 A와 사실상 동일하나 지후의 흰 동물뼈 목걸이가 사라졌고 전경 인물이 커져 '소프트한 좌하단 가장자리' 규정에서 멀어짐 — 계약이 요구하지 않은 변경으로 손실만 발생."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "프레이밍·방향·물리 모두 계약대로지만, 목/가슴이 프레임에 들어와 있음에도 지후의 흰 동물뼈 목걸이(CARRIED STATE 지속 조항)가 누락되어 유지 계약을 어겼다.  ★위반: [gpt] 우측 어선 선체에 검은 문자형 표식이 보여 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "동일한 와이드 트래킹 구도·마을 방향 mid-stride(한 발 지면 접지)를 그대로 유지하면서, 계약이 명시한 흰 동물뼈 목걸이만 추가로 복원했고 인물 의상·배경·광질 등 다른 요소는 사실상 변경 없음 — 얻은 것이 비용보다 크다.  ★위반: [gpt] 우측 어선 선체에 검은 문자형 표식이 보여 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.611,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.361,
      "B": 1.75
     },
     "violations": {
      "A": [
       "[gpt] 우측 어선 선체에 검은 문자형 표식이 보여 ‘어디에도 텍스트 금지’ 조건을 위반한다."
      ],
      "B": [
       "[gpt] 우측 어선 선체에 검은 문자형 표식이 보여 ‘어디에도 텍스트 금지’ 조건을 위반한다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "프레이밍·방향·물리 모두 계약대로지만, 목/가슴이 프레임에 들어와 있음에도 지후의 흰 동물뼈 목걸이(CARRIED STATE 지속 조항)가 누락되어 유지 계약을 어겼다.  ★위반: [gpt] 우측 어선 선체에 검은 문자형 표식이 보여 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     },
     {
      "label": "A",
      "score": 9,
      "verdict_ko": "동일한 와이드 트래킹 구도·마을 방향 mid-stride(한 발 지면 접지)를 그대로 유지하면서, 계약이 명시한 흰 동물뼈 목걸이만 추가로 복원했고 인물 의상·배경·광질 등 다른 요소는 사실상 변경 없음 — 얻은 것이 비용보다 크다.  ★위반: [gpt] 우측 어선 선체에 검은 문자형 표식이 보여 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 17,
     "B": 13
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S20sh11"
  }
 },
 "S26sh2::bgfirst_bg": {
  "input_fingerprint": "0ac2f00becd3da8b",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 문틈으로 불빛이 새어 나오는 낡은 창고를 향해 고개를 돌린 채 시선을 고정한 김지후(강수리영)의 모습.\n\nLOCATION (lock): The restaurant's outdoor back garden, between its vegetable beds and a weathered shed leaking light through the door.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the chest-height pan in a three-quarter view of 김지후(강수리영), then hold briefly once her turned head and fixed gaze align with the worn storage shed in the right background. Place her in the left midground and preserve a clear diagonal sightline to the narrow line of light at the shed door gap.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, midground, looks toward storage shed door gap; storage shed in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: worn storage shed (door slightly open; light escaping through the gap) — The wall and door side containing the narrow gap face diagonally toward the camera; used as Destination of 김지후(강수리영)'s newly fixed attention; vegetable garden (well tended); used as Low environmental layer separating 김지후(강수리영) from the shed; wooden platform with drying vegetables (vegetables laid out to dry) — Its upper surface is visible at an oblique angle behind 김지후(강수리영); used as Adds a grounded yard element without blocking the sightline to the shed.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The night yard stays dark while the stated light escaping through the shed door gap forms the sole concentrated visual cue.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 문틈으로 불빛이 새어 나오는 낡은 창고를 향해 고개를 돌린 채 시선을 고정한 김지후(강수리영)의 모습.\n\nLOCATION (lock): The restaurant's outdoor back garden, between its vegetable beds and a weathered shed leaking light through the door.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the chest-height pan in a three-quarter view of 김지후(강수리영), then hold briefly once her turned head and fixed gaze align with the worn storage shed in the right background. Place her in the left midground and preserve a clear diagonal sightline to the narrow line of light at the shed door gap.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, midground, looks toward storage shed door gap; storage shed in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: worn storage shed (door slightly open; light escaping through the gap) — The wall and door side containing the narrow gap face diagonally toward the camera; used as Destination of 김지후(강수리영)'s newly fixed attention; vegetable garden (well tended); used as Low environmental layer separating 김지후(강수리영) from the shed; wooden platform with drying vegetables (vegetables laid out to dry) — Its upper surface is visible at an oblique angle behind 김지후(강수리영); used as Adds a grounded yard element without blocking the sightline to the shed.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The night yard stays dark while the stated light escaping through the shed door gap forms the sole concentrated visual cue.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S26sh2__bgfirst_bg.png",
  "asset_id": "9e48ae8e-4a58-42b6-8f18-9e67de07c728",
  "input_asset_ids": [
   "7b1afcea-3ec9-4684-b3b3-747f9f334bda",
   "f131d46b-785b-495d-b829-b6b6f8eabc8f",
   "a724fb27-abad-4d06-bbe9-79bbb37368ab"
  ]
 },
 "S26sh2": {
  "input_fingerprint": "6eac242be48aab50",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 문틈으로 불빛이 새어 나오는 낡은 창고를 향해 고개를 돌린 채 시선을 고정한 김지후(강수리영)의 모습.\n\nLOCATION (lock): The restaurant's outdoor back garden, between its vegetable beds and a weathered shed leaking light through the door. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the chest-height pan in a three-quarter view of 김지후(강수리영), then hold briefly once her turned head and fixed gaze align with the worn storage shed in the right background. Place her in the left midground and preserve a clear diagonal sightline to the narrow line of light at the shed door gap.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, midground, looks toward storage shed door gap; storage shed in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: worn storage shed (door slightly open; light escaping through the gap) — The wall and door side containing the narrow gap face diagonally toward the camera; used as Destination of 김지후(강수리영)'s newly fixed attention; vegetable garden (well tended); used as Low environmental layer separating 김지후(강수리영) from the shed; wooden platform with drying vegetables (vegetables laid out to dry) — Its upper surface is visible at an oblique angle behind 김지후(강수리영); used as Adds a grounded yard element without blocking the sightline to the shed.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The night yard stays dark while the stated light escaping through the shed door gap forms the sole concentrated visual cue.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears the animal-bone necklace as she notices the lit shed.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 문틈으로 불빛이 새어 나오는 낡은 창고를 향해 고개를 돌린 채 시선을 고정한 김지후(강수리영)의 모습.\n\nLOCATION (lock): The restaurant's outdoor back garden, between its vegetable beds and a weathered shed leaking light through the door. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the chest-height pan in a three-quarter view of 김지후(강수리영), then hold briefly once her turned head and fixed gaze align with the worn storage shed in the right background. Place her in the left midground and preserve a clear diagonal sightline to the narrow line of light at the shed door gap.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, midground, looks toward storage shed door gap; storage shed in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: worn storage shed (door slightly open; light escaping through the gap) — The wall and door side containing the narrow gap face diagonally toward the camera; used as Destination of 김지후(강수리영)'s newly fixed attention; vegetable garden (well tended); used as Low environmental layer separating 김지후(강수리영) from the shed; wooden platform with drying vegetables (vegetables laid out to dry) — Its upper surface is visible at an oblique angle behind 김지후(강수리영); used as Adds a grounded yard element without blocking the sightline to the shed.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The night yard stays dark while the stated light escaping through the shed door gap forms the sole concentrated visual cue.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears the animal-bone necklace as she notices the lit shed.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 문틈으로 불빛이 새어 나오는 낡은 창고를 향해 고개를 돌린 채 시선을 고정한 김지후(강수리영)의 모습.\n\nLOCATION (lock): The restaurant's outdoor back garden, between its vegetable beds and a weathered shed leaking light through the door. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the chest-height pan in a three-quarter view of 김지후(강수리영), then hold briefly once her turned head and fixed gaze align with the worn storage shed in the right background. Place her in the left midground and preserve a clear diagonal sightline to the narrow line of light at the shed door gap.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 김지후(강수리영) in the middle-left of the frame, midground, looks toward storage shed door gap; storage shed in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: worn storage shed (door slightly open; light escaping through the gap) — The wall and door side containing the narrow gap face diagonally toward the camera; used as Destination of 김지후(강수리영)'s newly fixed attention; vegetable garden (well tended); used as Low environmental layer separating 김지후(강수리영) from the shed; wooden platform with drying vegetables (vegetables laid out to dry) — Its upper surface is visible at an oblique angle behind 김지후(강수리영); used as Adds a grounded yard element without blocking the sightline to the shed.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The night yard stays dark while the stated light escaping through the shed door gap forms the sole concentrated visual cue.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears the animal-bone necklace as she notices the lit shed.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S26sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S26sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:813104>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L22B03.png"
    },
    {
     "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_wooden_restaurant_compound_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:813104>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "all_candidates_fail": false,
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 좌측 중경에서 등을 반쯤 보인 채 고개를 오른쪽으로 돌려 우측 중경의 창고 쪽을 향한다. 시선 축은 대체로 창고 문틈의 세로 빛줄기와 맞으나, 인물 얼굴이 프레임 좌측 상단을 크게 차지해 대각선 시선 라인이 다소 눌린다. 창고 문은 살짝 열려 우측 가장자리에서 따뜻한 세로 빛이 새어나오고 있어 목표물은 명확히 식별된다.",
      "built_space": "블록조 창고(콘크리트 블록 벽 + 골함석 지붕)는 로케이션 사진의 구조와 일치하며 문 한 짝이 열려 안쪽으로 어둠, 문 옆 틈에서 빛. 우측에 본채 벽·격자창·툇마루·디딤돌이 로케이션 사진대로 배치. 좌측에 무말랭이 건조대와 채소밭이 있다. 다만 프롬프트가 지정한 '말린 채소가 놓인 나무 평상'은 보이지 않는다. 구조물 개수·개구부는 중복 없이 정상.",
      "entities": "인물은 20대 초반 한국 여성, 긴 검은 생머리, 캐릭터 레퍼런스 얼굴과 유사. 그러나 의상이 레퍼런스와 크게 어긋난다 — 모자 없음, 브라운 재킷 없음, 반팔 티셔츠 차림. 결정적으로 CARRIED STATE로 명시된 동물 뼈 목걸이가 전혀 보이지 않는다(목 부분이 프레임 안에 있음에도 부재). 창고는 야간 조명 하 실제 재질감으로 읽힘.",
      "hard_violations": [],
      "physics": "인물은 두 발이 프레임 밖이지만 흉부 이상 상체가 자연스러운 무게중심으로 서 있고, 지면과의 관계가 어긋나지 않음. 공중에 뜬 물체 없음. 빛은 문틈에서 실제 광원처럼 확산되나 바닥으로의 빛 번짐은 약하다."
     },
     {
      "label": "B",
      "direction": "김지후는 화면 좌측 중경에 서서 몸과 얼굴을 우측으로 돌려 우측 배경의 창고 문틈을 정면으로 응시한다. 시선 벡터가 창고 문 우측의 세로 빛줄기와 정확히 일치하고, 그 사이를 채소밭이라는 낮은 레이어가 가로막지 않는 명확한 대각선 사선이 형성된다.",
      "built_space": "블록조 창고, 두 짝 나무문, 우측 문틈에서 세로로 새는 따뜻한 빛, 골함석 지붕 — 로케이션 사진 구조와 일치. 좌측에 무말랭이 건조대, 마늘/채소 이랑, 그리고 프롬프트가 요구한 '말린 채소가 놓인 나무 평상'이 인물 뒤 비스듬한 각도로 존재. 배경 콘크리트 담장도 로케이션과 일치. 구조 중복 없음.",
      "entities": "20대 초반 한국 여성, 검은 머리(뒤로 묶음 — 레퍼런스의 완전한 긴 생머리와는 약간 다름), 캐릭터 레퍼런스의 브라운 워크재킷·올리브 이너·다크 카고팬츠·워크부츠를 정확히 착용. 목에는 흰 동물 뼈(가지 모양) 펜던트 목걸이가 선명히 보여 CARRIED STATE 충족. 모자는 미착용.",
      "hard_violations": [],
      "physics": "인물은 두 발 모두 흙바닥을 딛고 체중을 왼발 쪽에 실은 자연스러운 정지 자세, 팔은 늘어뜨렸으나 어깨 회전으로 살아있는 순간감을 유지. 공중에 뜬 요소 없음. 문틈 빛이 바닥에 실제로 길쭉한 광선 패치를 만들어 광학적으로 일관됨."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "시선 방향과 창고 문틈 빛은 맞지만 프레이밍이 미디엄샷보다 훨씬 타이트해 인물이 근경 클로즈업처럼 커졌고, 캐릭터 레퍼런스 의상(재킷·모자)과 필수 지속 소품인 동물 뼈 목걸이가 누락됐으며 나무 평상도 없다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "미디엄샷 스케일, 좌중간 인물/우중간 창고 배치, 문틈 빛으로의 명확한 대각선 시선, 채소밭·건조 채소 평상 등 지정 배경 요소, 그리고 동물 뼈 목걸이와 레퍼런스 의상까지 모두 충족한다(머리 묶음·모자 부재만 소소한 차이)."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "시선 방향과 창고 문틈 빛은 맞지만 프레이밍이 미디엄샷보다 훨씬 타이트해 인물이 근경 클로즈업처럼 커졌고, 캐릭터 레퍼런스 의상(재킷·모자)과 필수 지속 소품인 동물 뼈 목걸이가 누락됐으며 나무 평상도 없다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "미디엄샷 스케일, 좌중간 인물/우중간 창고 배치, 문틈 빛으로의 명확한 대각선 시선, 채소밭·건조 채소 평상 등 지정 배경 요소, 그리고 동물 뼈 목걸이와 레퍼런스 의상까지 모두 충족한다(머리 묶음·모자 부재만 소소한 차이)."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 좌측 중경에서 등을 반쯤 보인 채 고개를 오른쪽으로 돌려 우측 중경의 창고 쪽을 향한다. 시선 축은 대체로 창고 문틈의 세로 빛줄기와 맞으나, 인물 얼굴이 프레임 좌측 상단을 크게 차지해 대각선 시선 라인이 다소 눌린다. 창고 문은 살짝 열려 우측 가장자리에서 따뜻한 세로 빛이 새어나오고 있어 목표물은 명확히 식별된다.",
      "built_space": "블록조 창고(콘크리트 블록 벽 + 골함석 지붕)는 로케이션 사진의 구조와 일치하며 문 한 짝이 열려 안쪽으로 어둠, 문 옆 틈에서 빛. 우측에 본채 벽·격자창·툇마루·디딤돌이 로케이션 사진대로 배치. 좌측에 무말랭이 건조대와 채소밭이 있다. 다만 프롬프트가 지정한 '말린 채소가 놓인 나무 평상'은 보이지 않는다. 구조물 개수·개구부는 중복 없이 정상.",
      "entities": "인물은 20대 초반 한국 여성, 긴 검은 생머리, 캐릭터 레퍼런스 얼굴과 유사. 그러나 의상이 레퍼런스와 크게 어긋난다 — 모자 없음, 브라운 재킷 없음, 반팔 티셔츠 차림. 결정적으로 CARRIED STATE로 명시된 동물 뼈 목걸이가 전혀 보이지 않는다(목 부분이 프레임 안에 있음에도 부재). 창고는 야간 조명 하 실제 재질감으로 읽힘.",
      "hard_violations": [],
      "physics": "인물은 두 발이 프레임 밖이지만 흉부 이상 상체가 자연스러운 무게중심으로 서 있고, 지면과의 관계가 어긋나지 않음. 공중에 뜬 물체 없음. 빛은 문틈에서 실제 광원처럼 확산되나 바닥으로의 빛 번짐은 약하다."
     },
     {
      "label": "B",
      "direction": "김지후는 화면 좌측 중경에 서서 몸과 얼굴을 우측으로 돌려 우측 배경의 창고 문틈을 정면으로 응시한다. 시선 벡터가 창고 문 우측의 세로 빛줄기와 정확히 일치하고, 그 사이를 채소밭이라는 낮은 레이어가 가로막지 않는 명확한 대각선 사선이 형성된다.",
      "built_space": "블록조 창고, 두 짝 나무문, 우측 문틈에서 세로로 새는 따뜻한 빛, 골함석 지붕 — 로케이션 사진 구조와 일치. 좌측에 무말랭이 건조대, 마늘/채소 이랑, 그리고 프롬프트가 요구한 '말린 채소가 놓인 나무 평상'이 인물 뒤 비스듬한 각도로 존재. 배경 콘크리트 담장도 로케이션과 일치. 구조 중복 없음.",
      "entities": "20대 초반 한국 여성, 검은 머리(뒤로 묶음 — 레퍼런스의 완전한 긴 생머리와는 약간 다름), 캐릭터 레퍼런스의 브라운 워크재킷·올리브 이너·다크 카고팬츠·워크부츠를 정확히 착용. 목에는 흰 동물 뼈(가지 모양) 펜던트 목걸이가 선명히 보여 CARRIED STATE 충족. 모자는 미착용.",
      "hard_violations": [],
      "physics": "인물은 두 발 모두 흙바닥을 딛고 체중을 왼발 쪽에 실은 자연스러운 정지 자세, 팔은 늘어뜨렸으나 어깨 회전으로 살아있는 순간감을 유지. 공중에 뜬 요소 없음. 문틈 빛이 바닥에 실제로 길쭉한 광선 패치를 만들어 광학적으로 일관됨."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 프레임 좌측 중경에 3/4 측면으로 서서 고개를 오른쪽으로 돌려 시선이 우측 배경의 창고 문틈(세로로 새어나오는 빛줄기)에 정확히 떨어진다. 시선 라인이 밭이랑 위를 지나 문틈까지 막힘 없이 대각으로 이어진다.",
      "built_space": "창고는 우측 배경, 시멘트 블록 벽 + 골강판 지붕, 나무 여닫이 문이 살짝 열려 세로 빛줄기와 바닥으로 번지는 빛을 만든다. 로케이션 사진의 블록 창고 구조/재질과 일치. 좌측에는 무·시래기를 널어놓은 건조대와 파밭, 앞쪽에 나무 평상(채소 널린 상판이 비스듬히 보임)이 배치되어 계약된 배경 요소 3종이 모두 존재. 창고 크기는 인물 대비 실제 스케일에 부합.",
      "entities": "인물은 20대 초반 한국 여성, 긴 검은 머리, 갈색 워크재킷·올리브 헨리·짙은 카고팬츠·워크부츠로 캐릭터 레퍼런스 의상과 일치. 목에 뼈 모양(사슴뿔) 펜던트 목걸이가 명확히 보여 CARRIED STATE 충족. 눈은 정상 인체 눈. 텍스트·워터마크 없음.",
      "hard_violations": [],
      "physics": "두 발이 마당 흙바닥을 딛고 서 있으며 체중은 약간 왼쪽 다리에 실린 자연스러운 정지 자세. 팔은 몸 옆으로 편안히 늘어져 있고 공중에 뜬 물체 없음. 건조대·평상 모두 다리로 지지됨."
     },
     {
      "label": "B",
      "direction": "인물은 카메라 쪽으로 등을 보인 채 고개를 오른쪽으로 돌려 배경 중앙의 창고 문틈 빛을 바라본다. 시선은 창고 문 틈의 세로 빛줄기에 대체로 맞는다. 다만 인물이 화면 중앙-좌하단에 크게 자리해 프레임을 지배한다.",
      "built_space": "창고는 화면 중앙에 블록 벽·골강판 지붕, 문이 열려 세로 빛줄기가 나온다. 오른쪽에 로케이션 사진의 본채(흰 벽, 나무 창호, 툇마루, 디딤돌)가 재현되어 장소 일치도는 높다. 좌측에 시래기 건조대와 밭이 있으나 '채소를 널어놓은 나무 평상'은 보이지 않는다.",
      "entities": "20대 한국 여성, 긴 검은 머리는 맞으나 의상이 캐릭터 레퍼런스와 불일치 — 갈색 워크재킷과 카고팬츠 대신 반팔 티셔츠 차림. 목의 동물뼈 목걸이가 전혀 보이지 않아 CARRIED STATE 미충족(등/측면 프레임이지만 앞가슴 라인이 보이는 각도에서도 부재). 눈은 정상.",
      "hard_violations": [],
      "physics": "인물은 지면에 서 있는 것으로 보이나 하반신이 프레임 밖(가슴~허리 아래 크롭)이라 지지면은 추정. 공중에 뜬 요소 없음. 배경 구조물은 지면에 안착."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 9,
      "verdict_ko": "미디엄 샷·좌중경 3/4 자세·창고 문틈으로 이어지는 대각 시선·밭/건조 평상 배경 요소·동물뼈 목걸이·야간 단일 광원까지 지시된 프레임 계약을 거의 모두 충족한다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "장소 재현과 시선 방향은 맞지만 뒤통수 위주의 근접 크롭으로 지정된 '중경 3/4 미디엄 샷'을 벗어났고, 의상이 레퍼런스와 다르며 동물뼈 목걸이와 채소 널린 나무 평상이 보이지 않는다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "미디엄 샷·좌중경 3/4 자세·창고 문틈으로 이어지는 대각 시선·밭/건조 평상 배경 요소·동물뼈 목걸이·야간 단일 광원까지 지시된 프레임 계약을 거의 모두 충족한다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "장소 재현과 시선 방향은 맞지만 뒤통수 위주의 근접 크롭으로 지정된 '중경 3/4 미디엄 샷'을 벗어났고, 의상이 레퍼런스와 다르며 동물뼈 목걸이와 채소 널린 나무 평상이 보이지 않는다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "김지후는 프레임 좌측 중경에 3/4 측면으로 서서 고개를 오른쪽으로 돌려 시선이 우측 배경의 창고 문틈(세로로 새어나오는 빛줄기)에 정확히 떨어진다. 시선 라인이 밭이랑 위를 지나 문틈까지 막힘 없이 대각으로 이어진다.",
      "built_space": "창고는 우측 배경, 시멘트 블록 벽 + 골강판 지붕, 나무 여닫이 문이 살짝 열려 세로 빛줄기와 바닥으로 번지는 빛을 만든다. 로케이션 사진의 블록 창고 구조/재질과 일치. 좌측에는 무·시래기를 널어놓은 건조대와 파밭, 앞쪽에 나무 평상(채소 널린 상판이 비스듬히 보임)이 배치되어 계약된 배경 요소 3종이 모두 존재. 창고 크기는 인물 대비 실제 스케일에 부합.",
      "entities": "인물은 20대 초반 한국 여성, 긴 검은 머리, 갈색 워크재킷·올리브 헨리·짙은 카고팬츠·워크부츠로 캐릭터 레퍼런스 의상과 일치. 목에 뼈 모양(사슴뿔) 펜던트 목걸이가 명확히 보여 CARRIED STATE 충족. 눈은 정상 인체 눈. 텍스트·워터마크 없음.",
      "hard_violations": [],
      "physics": "두 발이 마당 흙바닥을 딛고 서 있으며 체중은 약간 왼쪽 다리에 실린 자연스러운 정지 자세. 팔은 몸 옆으로 편안히 늘어져 있고 공중에 뜬 물체 없음. 건조대·평상 모두 다리로 지지됨."
     },
     {
      "label": "A",
      "direction": "인물은 카메라 쪽으로 등을 보인 채 고개를 오른쪽으로 돌려 배경 중앙의 창고 문틈 빛을 바라본다. 시선은 창고 문 틈의 세로 빛줄기에 대체로 맞는다. 다만 인물이 화면 중앙-좌하단에 크게 자리해 프레임을 지배한다.",
      "built_space": "창고는 화면 중앙에 블록 벽·골강판 지붕, 문이 열려 세로 빛줄기가 나온다. 오른쪽에 로케이션 사진의 본채(흰 벽, 나무 창호, 툇마루, 디딤돌)가 재현되어 장소 일치도는 높다. 좌측에 시래기 건조대와 밭이 있으나 '채소를 널어놓은 나무 평상'은 보이지 않는다.",
      "entities": "20대 한국 여성, 긴 검은 머리는 맞으나 의상이 캐릭터 레퍼런스와 불일치 — 갈색 워크재킷과 카고팬츠 대신 반팔 티셔츠 차림. 목의 동물뼈 목걸이가 전혀 보이지 않아 CARRIED STATE 미충족(등/측면 프레임이지만 앞가슴 라인이 보이는 각도에서도 부재). 눈은 정상.",
      "hard_violations": [],
      "physics": "인물은 지면에 서 있는 것으로 보이나 하반신이 프레임 밖(가슴~허리 아래 크롭)이라 지지면은 추정. 공중에 뜬 요소 없음. 배경 구조물은 지면에 안착."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 10,
     "B": 17
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "readings": [
   {
    "label": "A",
    "direction": "김지후는 좌측 중경에서 등을 반쯤 보인 채 고개를 오른쪽으로 돌려 우측 중경의 창고 쪽을 향한다. 시선 축은 대체로 창고 문틈의 세로 빛줄기와 맞으나, 인물 얼굴이 프레임 좌측 상단을 크게 차지해 대각선 시선 라인이 다소 눌린다. 창고 문은 살짝 열려 우측 가장자리에서 따뜻한 세로 빛이 새어나오고 있어 목표물은 명확히 식별된다.",
    "built_space": "블록조 창고(콘크리트 블록 벽 + 골함석 지붕)는 로케이션 사진의 구조와 일치하며 문 한 짝이 열려 안쪽으로 어둠, 문 옆 틈에서 빛. 우측에 본채 벽·격자창·툇마루·디딤돌이 로케이션 사진대로 배치. 좌측에 무말랭이 건조대와 채소밭이 있다. 다만 프롬프트가 지정한 '말린 채소가 놓인 나무 평상'은 보이지 않는다. 구조물 개수·개구부는 중복 없이 정상.",
    "entities": "인물은 20대 초반 한국 여성, 긴 검은 생머리, 캐릭터 레퍼런스 얼굴과 유사. 그러나 의상이 레퍼런스와 크게 어긋난다 — 모자 없음, 브라운 재킷 없음, 반팔 티셔츠 차림. 결정적으로 CARRIED STATE로 명시된 동물 뼈 목걸이가 전혀 보이지 않는다(목 부분이 프레임 안에 있음에도 부재). 창고는 야간 조명 하 실제 재질감으로 읽힘.",
    "hard_violations": [],
    "physics": "인물은 두 발이 프레임 밖이지만 흉부 이상 상체가 자연스러운 무게중심으로 서 있고, 지면과의 관계가 어긋나지 않음. 공중에 뜬 물체 없음. 빛은 문틈에서 실제 광원처럼 확산되나 바닥으로의 빛 번짐은 약하다."
   },
   {
    "label": "B",
    "direction": "김지후는 화면 좌측 중경에 서서 몸과 얼굴을 우측으로 돌려 우측 배경의 창고 문틈을 정면으로 응시한다. 시선 벡터가 창고 문 우측의 세로 빛줄기와 정확히 일치하고, 그 사이를 채소밭이라는 낮은 레이어가 가로막지 않는 명확한 대각선 사선이 형성된다.",
    "built_space": "블록조 창고, 두 짝 나무문, 우측 문틈에서 세로로 새는 따뜻한 빛, 골함석 지붕 — 로케이션 사진 구조와 일치. 좌측에 무말랭이 건조대, 마늘/채소 이랑, 그리고 프롬프트가 요구한 '말린 채소가 놓인 나무 평상'이 인물 뒤 비스듬한 각도로 존재. 배경 콘크리트 담장도 로케이션과 일치. 구조 중복 없음.",
    "entities": "20대 초반 한국 여성, 검은 머리(뒤로 묶음 — 레퍼런스의 완전한 긴 생머리와는 약간 다름), 캐릭터 레퍼런스의 브라운 워크재킷·올리브 이너·다크 카고팬츠·워크부츠를 정확히 착용. 목에는 흰 동물 뼈(가지 모양) 펜던트 목걸이가 선명히 보여 CARRIED STATE 충족. 모자는 미착용.",
    "hard_violations": [],
    "physics": "인물은 두 발 모두 흙바닥을 딛고 체중을 왼발 쪽에 실은 자연스러운 정지 자세, 팔은 늘어뜨렸으나 어깨 회전으로 살아있는 순간감을 유지. 공중에 뜬 요소 없음. 문틈 빛이 바닥에 실제로 길쭉한 광선 패치를 만들어 광학적으로 일관됨."
   }
  ],
  "totals": {
   "A": 10,
   "B": 17
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "시선 방향과 창고 문틈 빛은 맞지만 프레이밍이 미디엄샷보다 훨씬 타이트해 인물이 근경 클로즈업처럼 커졌고, 캐릭터 레퍼런스 의상(재킷·모자)과 필수 지속 소품인 동물 뼈 목걸이가 누락됐으며 나무 평상도 없다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "미디엄샷 스케일, 좌중간 인물/우중간 창고 배치, 문틈 빛으로의 명확한 대각선 시선, 채소밭·건조 채소 평상 등 지정 배경 요소, 그리고 동물 뼈 목걸이와 레퍼런스 의상까지 모두 충족한다(머리 묶음·모자 부재만 소소한 차이)."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L22B03.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_wooden_restaurant_compound_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:813104>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트에서 미디엄 샷(medium shot)을 요구했으나, 인물의 전신과 발까지 모두 나오는 풀 샷(full shot)으로 생성되었습니다.",
     "fix_en": "Requires regeneration to change the camera distance. The image must be shot as a medium shot framing Jihu from the waist up.",
     "needs_regeneration": true
    },
    {
     "issue_ko": "창고 건물이 지정된 STRUCTURE LOOK 레퍼런스(하얀 목재 외벽)를 무시하고, LOCATION 레퍼런스의 시멘트 블록 재질과 임의의 양여닫이문으로 잘못 렌더링되었습니다.",
     "fix_en": "Change the shed's walls to white painted vertical wooden planks and adjust the door style to match the structure look reference. Preserve Jihu's pose, wardrobe, the garden layout, lighting, and camera framing exactly.",
     "needs_regeneration": true
    },
    {
     "issue_ko": "프롬프트상 인물 뒤편(behind 김지후)에 위치해야 할 채소 건조 평상이 인물 앞쪽 전경(좌측 하단)에 배치되어 공간 배치가 어긋났습니다.",
     "fix_en": "Remove the wooden platform from the foreground and place it in the midground behind Jihu. Preserve Jihu's pose, wardrobe, the shed, garden plants, lighting, and camera framing exactly.",
     "needs_regeneration": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Requires regeneration to change the camera distance. The image must be shot as a medium shot framing Jihu from the waist up.\n- Change the shed's walls to white painted vertical wooden planks and adjust the door style to match the structure look reference. Preserve Jihu's pose, wardrobe, the garden layout, lighting, and camera framing exactly.\n- Remove the wooden platform from the foreground and place it in the midground behind Jihu. Preserve Jihu's pose, wardrobe, the shed, garden plants, lighting, and camera framing exactly.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "미디엄 샷·중좌측 미드그라운드 인물·우측 배경 창고·문틈 한 줄 빛이라는 프레임 지시를 정확히 지켰고, 창고의 시멘트 블록 재질과 평상 위 말리는 채소 등 로케이션 요소가 일치하며 배경 스케일도 실제 크기 그대로다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "시선·인물·목걸이는 동일하나 계약이 요구하지 않은 변경(창고를 블록조에서 흰 판벽으로 교체, 창고 확대로 배경이 전경화, 평상 위 말리는 채소 삭제 후 텃밭에 선반 신설, 문 전체가 밝아져 '문틈 빛' 단서 약화)이 누적되어 손실이 이득보다 크다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.857,
      "B": 1.75
     },
     "adjusted": {
      "A": 1.857,
      "B": 1.75
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "B"
     },
     "agreed": false
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "미디엄 샷·중좌측 미드그라운드 인물·우측 배경 창고·문틈 한 줄 빛이라는 프레임 지시를 정확히 지켰고, 창고의 시멘트 블록 재질과 평상 위 말리는 채소 등 로케이션 요소가 일치하며 배경 스케일도 실제 크기 그대로다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "시선·인물·목걸이는 동일하나 계약이 요구하지 않은 변경(창고를 블록조에서 흰 판벽으로 교체, 창고 확대로 배경이 전경화, 평상 위 말리는 채소 삭제 후 텃밭에 선반 신설, 문 전체가 밝아져 '문틈 빛' 단서 약화)이 누적되어 손실이 이득보다 크다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "인물·시선·구도는 맞지만 창고 벽체를 흰 목재판으로 바꿔 LOCATION/STRUCTURE의 시멘트 블록 재질 락을 어겼고, 문 전체가 발광해 '문틈으로 새어 나오는 좁은 빛'이라는 샷 텍스트와 조명 계약을 약화시켰으며 창고가 과장되게 커졌다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "시멘트 블록 창고 재질과 크기 비례가 레퍼런스와 일치하고, 문틈에서만 새어 나오는 가느다란 빛이 유일한 광원 큐로 작동하며, 좌중경 3/4 미디엄샷·평상 위 말리는 채소·목걸이 등 계약 항목을 그대로 충족한다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.464,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.464,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "인물·시선·구도는 맞지만 창고 벽체를 흰 목재판으로 바꿔 LOCATION/STRUCTURE의 시멘트 블록 재질 락을 어겼고, 문 전체가 발광해 '문틈으로 새어 나오는 좁은 빛'이라는 샷 텍스트와 조명 계약을 약화시켰으며 창고가 과장되게 커졌다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "시멘트 블록 창고 재질과 크기 비례가 레퍼런스와 일치하고, 문틈에서만 새어 나오는 가느다란 빛이 유일한 광원 큐로 작동하며, 좌중경 3/4 미디엄샷·평상 위 말리는 채소·목걸이 등 계약 항목을 그대로 충족한다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 12
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S26sh2__bgfirst_bg.png",
   "bg_asset_id": "9e48ae8e-4a58-42b6-8f18-9e67de07c728",
   "bg_record_key": "S26sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate",
   "seed_attached": true
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "groupbg::해송 숲길": {
  "input_fingerprint": "26ba5c19ec3d878b",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "해송 숲길",
    "tags": [
     "S107sh3",
     "S23sh1",
     "S29sh2",
     "S29sh5",
     "S30sh2",
     "S30sh4",
     "S30sh7",
     "S45sh10",
     "S45sh3",
     "S45sh9",
     "S64sh3",
     "S67sh3",
     "S67sh8",
     "S67sh9",
     "S69sh4",
     "S69sh9",
     "S74sh2",
     "S74sh8"
    ]
   },
   "context_sig": "d154e24b7344158f"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated: A narrow nighttime trail deep in the island's black-pine forest.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 해송 숲길과 산길: 하늘을 가릴 듯 울창한 흑빛 소나무(해송)가 빽빽하게 찬 어둡고 좁은 흙길. 얽힌 수풀과 잔가지가 통행을 방해한다. 곳곳에 이끼가 잔뜩 낀 무릎 높이의 돌 비석들이 세워져 있고, 그 앞 바닥에는 핏물이 고인 썩은 고깃덩어리들이 놓여 있다. 밤에는 달빛만 간신히 비친다. (특징: 빽빽한 검은빛 소나무 군락; 어두운 나무 그늘과 무성한 수풀; 좁은 흙길 오솔길; 이끼 낀 돌 비석들; 비석 앞의 부패한 붉은 고깃덩어리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 어슴프레 달빛만이 비치는 울창한 해송 나무 숲을 지나가는 산들.\n- 핸드폰 플래쉬 불빛이 어둑한 해송 나무 숲을 비춘다.\n- 영란이 산책하듯 가벼운 발걸음으로 숲길을 내려온다.\n- 달빛 아래 헉헉대며 숲길을 올라가는 수리영과 인우.\n- 실외. 숲 속 - 밤\n\nTIME OF DAY (lock): night.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated: A narrow nighttime trail deep in the island's black-pine forest.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 해송 숲길과 산길: 하늘을 가릴 듯 울창한 흑빛 소나무(해송)가 빽빽하게 찬 어둡고 좁은 흙길. 얽힌 수풀과 잔가지가 통행을 방해한다. 곳곳에 이끼가 잔뜩 낀 무릎 높이의 돌 비석들이 세워져 있고, 그 앞 바닥에는 핏물이 고인 썩은 고깃덩어리들이 놓여 있다. 밤에는 달빛만 간신히 비친다. (특징: 빽빽한 검은빛 소나무 군락; 어두운 나무 그늘과 무성한 수풀; 좁은 흙길 오솔길; 이끼 낀 돌 비석들; 비석 앞의 부패한 붉은 고깃덩어리)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 어슴프레 달빛만이 비치는 울창한 해송 나무 숲을 지나가는 산들.\n- 핸드폰 플래쉬 불빛이 어둑한 해송 나무 숲을 비춘다.\n- 영란이 산책하듯 가벼운 발걸음으로 숲길을 내려온다.\n- 달빛 아래 헉헉대며 숲길을 올라가는 수리영과 인우.\n- 실외. 숲 속 - 밤\n\nTIME OF DAY (lock): night.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송_숲길_cf7470.png",
  "asset_id": "26370ef0-070a-470c-beac-42f57a83eef4",
  "input_asset_ids": [
   "a4214bf0-5fc6-4a4f-be84-af37e359c589"
  ],
  "origin_tag": "S29sh2",
  "place_text": "A narrow nighttime trail deep in the island's black-pine forest.",
  "origin_inputs": {
   "place_text": "A narrow nighttime trail deep in the island's black-pine forest.",
   "time_of_day_en": "night",
   "conti_asset_id": "a4214bf0-5fc6-4a4f-be84-af37e359c589"
  }
 },
 "S29sh2::bgfirst_bg": {
  "input_fingerprint": "093ce919456e1bf5",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어둠 속 숲길에서 금월도 마을 청년들이 무언가 꿈틀거리는 커다란 포댓자루 두 개를 실은 리어카 손잡이를 쥔 채, 몸을 앞으로 기울여 한쪽 발을 내딛은 mid-action 광경.\n\nLOCATION (lock): A narrow nighttime trail deep in the island's black-pine forest.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the flow’s slightly lowered chest-height position beside the concealed observer, pan across the path in a wide lateral reveal as two village youths lean into the cart handles and step through the frame. Keep their hands and forward-pitched torsos in the middle plane, with the cart and both visibly moving sacks fully readable behind them without letting either sack occupy more than a third of the image.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: leading village youth in the middle-right of the frame, midground, moves toward route deeper into the forest; second village youth in the middle-center of the frame, midground, moves toward route deeper into the forest; route deeper into the forest in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 좁은 숲길 (Surrounded by 해송 trees) — The path crosses the frame and continues toward screen right; used as Provides the narrow lateral route across which the group is revealed; 리어카 (Loaded with two large sacks) — Its side and both handles are visible as it advances across the camera axis; used as Carries the two sacks and connects the youths' grips to their forward movement; 커다란 포댓자루 두 개 (Loaded on the cart; contents moving) — Their outward-facing sides show intermittent bulges from movement inside; used as Both remain visible behind the youths, with localized movement readable in their outlines.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Very low nighttime visibility in the forest after 김지후(강수리영) has switched off her phone light.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어둠 속 숲길에서 금월도 마을 청년들이 무언가 꿈틀거리는 커다란 포댓자루 두 개를 실은 리어카 손잡이를 쥔 채, 몸을 앞으로 기울여 한쪽 발을 내딛은 mid-action 광경.\n\nLOCATION (lock): A narrow nighttime trail deep in the island's black-pine forest.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the flow’s slightly lowered chest-height position beside the concealed observer, pan across the path in a wide lateral reveal as two village youths lean into the cart handles and step through the frame. Keep their hands and forward-pitched torsos in the middle plane, with the cart and both visibly moving sacks fully readable behind them without letting either sack occupy more than a third of the image.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: leading village youth in the middle-right of the frame, midground, moves toward route deeper into the forest; second village youth in the middle-center of the frame, midground, moves toward route deeper into the forest; route deeper into the forest in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 좁은 숲길 (Surrounded by 해송 trees) — The path crosses the frame and continues toward screen right; used as Provides the narrow lateral route across which the group is revealed; 리어카 (Loaded with two large sacks) — Its side and both handles are visible as it advances across the camera axis; used as Carries the two sacks and connects the youths' grips to their forward movement; 커다란 포댓자루 두 개 (Loaded on the cart; contents moving) — Their outward-facing sides show intermittent bulges from movement inside; used as Both remain visible behind the youths, with localized movement readable in their outlines.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Very low nighttime visibility in the forest after 김지후(강수리영) has switched off her phone light.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S29sh2__bgfirst_bg.png",
  "asset_id": "8bb225ff-4300-4acc-8286-35008779b609",
  "input_asset_ids": [
   "a4214bf0-5fc6-4a4f-be84-af37e359c589",
   "26370ef0-070a-470c-beac-42f57a83eef4"
  ]
 },
 "S29sh2": {
  "input_fingerprint": "467755ff8e126198",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어둠 속 숲길에서 금월도 마을 청년들이 무언가 꿈틀거리는 커다란 포댓자루 두 개를 실은 리어카 손잡이를 쥔 채, 몸을 앞으로 기울여 한쪽 발을 내딛은 mid-action 광경.\n\nLOCATION (lock): A narrow nighttime trail deep in the island's black-pine forest. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the flow’s slightly lowered chest-height position beside the concealed observer, pan across the path in a wide lateral reveal as two village youths lean into the cart handles and step through the frame. Keep their hands and forward-pitched torsos in the middle plane, with the cart and both visibly moving sacks fully readable behind them without letting either sack occupy more than a third of the image.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: leading village youth in the middle-right of the frame, midground, moves toward route deeper into the forest; second village youth in the middle-center of the frame, midground, moves toward route deeper into the forest; route deeper into the forest in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 좁은 숲길 (Surrounded by 해송 trees) — The path crosses the frame and continues toward screen right; used as Provides the narrow lateral route across which the group is revealed; 리어카 (Loaded with two large sacks) — Its side and both handles are visible as it advances across the camera axis; used as Carries the two sacks and connects the youths' grips to their forward movement; 커다란 포댓자루 두 개 (Loaded on the cart; contents moving) — Their outward-facing sides show intermittent bulges from movement inside; used as Both remain visible behind the youths, with localized movement readable in their outlines.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Very low nighttime visibility in the forest after 김지후(강수리영) has switched off her phone light.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu continues to wear her animal-bone necklace while following the cart.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 금월도 마을 청년들 right now, so 금월도 마을 청년들's hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 금월도 마을 청년들: its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 마을 청년 #1 (Korean young man, short dark hair, broad youthful face); 마을 청년 #2 (Korean young man, short dark hair, lean youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어둠 속 숲길에서 금월도 마을 청년들이 무언가 꿈틀거리는 커다란 포댓자루 두 개를 실은 리어카 손잡이를 쥔 채, 몸을 앞으로 기울여 한쪽 발을 내딛은 mid-action 광경.\n\nLOCATION (lock): A narrow nighttime trail deep in the island's black-pine forest. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the flow’s slightly lowered chest-height position beside the concealed observer, pan across the path in a wide lateral reveal as two village youths lean into the cart handles and step through the frame. Keep their hands and forward-pitched torsos in the middle plane, with the cart and both visibly moving sacks fully readable behind them without letting either sack occupy more than a third of the image.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: leading village youth in the middle-right of the frame, midground, moves toward route deeper into the forest; second village youth in the middle-center of the frame, midground, moves toward route deeper into the forest; route deeper into the forest in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 좁은 숲길 (Surrounded by 해송 trees) — The path crosses the frame and continues toward screen right; used as Provides the narrow lateral route across which the group is revealed; 리어카 (Loaded with two large sacks) — Its side and both handles are visible as it advances across the camera axis; used as Carries the two sacks and connects the youths' grips to their forward movement; 커다란 포댓자루 두 개 (Loaded on the cart; contents moving) — Their outward-facing sides show intermittent bulges from movement inside; used as Both remain visible behind the youths, with localized movement readable in their outlines.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Very low nighttime visibility in the forest after 김지후(강수리영) has switched off her phone light.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu continues to wear her animal-bone necklace while following the cart.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 금월도 마을 청년들 right now, so 금월도 마을 청년들's hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 금월도 마을 청년들: its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 마을 청년 #1 (Korean young man, short dark hair, broad youthful face); 마을 청년 #2 (Korean young man, short dark hair, lean youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어둠 속 숲길에서 금월도 마을 청년들이 무언가 꿈틀거리는 커다란 포댓자루 두 개를 실은 리어카 손잡이를 쥔 채, 몸을 앞으로 기울여 한쪽 발을 내딛은 mid-action 광경.\n\nLOCATION (lock): A narrow nighttime trail deep in the island's black-pine forest. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the flow’s slightly lowered chest-height position beside the concealed observer, pan across the path in a wide lateral reveal as two village youths lean into the cart handles and step through the frame. Keep their hands and forward-pitched torsos in the middle plane, with the cart and both visibly moving sacks fully readable behind them without letting either sack occupy more than a third of the image.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: leading village youth in the middle-right of the frame, midground, moves toward route deeper into the forest; second village youth in the middle-center of the frame, midground, moves toward route deeper into the forest; route deeper into the forest in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 좁은 숲길 (Surrounded by 해송 trees) — The path crosses the frame and continues toward screen right; used as Provides the narrow lateral route across which the group is revealed; 리어카 (Loaded with two large sacks) — Its side and both handles are visible as it advances across the camera axis; used as Carries the two sacks and connects the youths' grips to their forward movement; 커다란 포댓자루 두 개 (Loaded on the cart; contents moving) — Their outward-facing sides show intermittent bulges from movement inside; used as Both remain visible behind the youths, with localized movement readable in their outlines.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Very low nighttime visibility in the forest after 김지후(강수리영) has switched off her phone light.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu continues to wear her animal-bone necklace while following the cart.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 금월도 마을 청년들 right now, so 금월도 마을 청년들's hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 금월도 마을 청년들: its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 마을 청년 #1 (Korean young man, short dark hair, broad youthful face); 마을 청년 #2 (Korean young man, short dark hair, lean youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S29sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S29sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 마을 청년 #1: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:976931>"
    },
    {
     "label": "CHARACTER REFERENCE — 마을 청년 #2: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1020350>"
    },
    {
     "label": "PROP REFERENCE — 피 묻은 포댓자루: the exact object appearing in this shot; match its look, material and wear exactly.",
     "path": "<bytes:1566892>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송_숲길_cf7470.png"
    },
    {
     "label": "CHARACTER REFERENCE — 마을 청년 #1: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:976931>"
    },
    {
     "label": "CHARACTER REFERENCE — 마을 청년 #2: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:1020350>"
    },
    {
     "label": "PROP REFERENCE — 피 묻은 포댓자루: the exact object appearing in this shot; match its look, material and wear exactly.",
     "path": "<bytes:1566892>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 청년 모두 화면 오른쪽(숲 깊은 쪽 길)을 향해 시선과 몸을 두고 있고, 리어카는 카메라 축을 가로질러 우측으로 전진한다. 청년 #2(체크셔츠)의 두 손은 리어카 손잡이 두 개를 실제로 쥐고 있으며 팔이 손잡이 방향과 일치한다. 청년 #1(누더기·두건)은 그 앞쪽에서 같은 방향으로 몸을 기울인다. 프레임 좌측 가장자리에는 관찰자(김지후 추정)의 팔·다리 일부만 잘려 들어와 있어 은신 관찰자 옆 시점이라는 설정과 부합한다.",
      "built_space": "리어카는 측면이 보이며 바퀴 2개, 짐칸, 손잡이 두 개가 모두 한 세트로 정확히 렌더링됨. 손잡이는 뒤쪽(좌측)으로 뻗어 뒤에서 미는 사람의 손에 닿는다. 장소는 레퍼런스와 같은 해송림 좁은 길, 좌측 열 지어 선 이끼 낀 돌들(선돌 열)까지 일치.",
      "entities": "청년 #2는 붉은 체크셔츠·데님·부츠로 레퍼런스와 일치. 청년 #1은 회색 두건과 누더기 상의로 레퍼런스 의상과 일치(얼굴은 어둡게 처리). 포댓자루 두 개는 마대 질감에 어두운 붉은 얼룩, 상단 끈 묶음까지 프롭 레퍼런스와 부합하며 각각 화면 1/3 미만 차지. 좌측 잘린 인물은 얼굴 없는 부분 프레임이라 추가 인물 신설로 보긴 어려우나 신원 확인은 불가.",
      "hard_violations": [],
      "physics": "두 청년 모두 한쪽 발이 지면에 확실히 닿아 있고 앞으로 기운 상반신·뻗은 뒷다리로 밀고 나가는 하중이 읽힌다. 리어카는 두 바퀴가 땅에 닿아 있고 손잡이는 손에 지지됨. 공중에 뜬 물체 없음."
     },
     {
      "label": "B",
      "direction": "우측 청년 #1은 화면 오른쪽을 보며 리어카 손잡이 하나를 쥐고 앞으로 기울어 있다. 그러나 두 번째 청년(체크셔츠)은 리어카에서 멀리 떨어져 뒤쪽 길 위에 홀로 서서 카메라 쪽(좌측 전방)을 향해 몸을 굽히고 있어 손잡이를 전혀 쥐고 있지 않다 — 샷 텍스트의 '두 청년이 손잡이를 쥔 채'와 어긋난다. 좌측 전경에는 뼈 목걸이를 한 여성 인물의 목·가슴이 크게 들어와 있다.",
      "built_space": "리어카는 프레임 우측 가장자리에 잘려 짐칸 일부와 바퀴 하나만 보이고, 손잡이는 한쪽만 확인됨(양쪽 손잡이 모두 보이라는 지시 미충족). 나무 수레 형태로 현대식 리어카 특징(공기 타이어·철제 프레임)이 약함. 장소의 선돌 열과 해송은 레퍼런스와 부합.",
      "entities": "청년 #2 체크셔츠·데님·부츠 일치, 청년 #1 두건·누더기 일치. 포댓자루는 4개 이상 보이며 '두 개'라는 지시와 불일치, 마대 질감·붉은 얼룩은 프롭과 유사. 좌측 전경 여성 인물(김지후로 보이는 목걸이 착용자)이 얼굴 없이 크게 들어와 있으나 인물 목록에는 두 청년만 허용되어 있다.",
      "hard_violations": [
       "샷 텍스트가 지정한 인물 외 제3의 인물(뼈 목걸이 착용 여성)이 전경에 상반신 규모로 등장",
       "포댓자루가 두 개가 아니라 네 개 이상으로 렌더링됨"
      ],
      "physics": "두 인물 모두 한 발이 땅에 닿아 있고 지지는 있으나, 체크셔츠 청년은 손에 아무것도 없이 허리만 굽힌 어정쩡한 자세로 리어카를 미는 동작이 성립하지 않는다. 수레는 바퀴로 지지됨."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "두 청년이 실제로 리어카 손잡이를 쥐고 앞으로 기울여 한 발을 내딛는 mid-action이 성립하고, 리어카 측면·양쪽 손잡이·포댓자루 두 개가 모두 각 1/3 미만으로 읽히는 와이드 구도와 야간 저조도 로케이션 일치가 우수하다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "체크셔츠 청년이 손잡이를 전혀 쥐지 않고 따로 떨어져 있어 핵심 동작이 무너졌고, 포댓자루가 두 개가 아닌 다수이며 전경에 허용되지 않은 제3의 인물이 크게 들어와 하드 위반이다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "두 청년이 실제로 리어카 손잡이를 쥐고 앞으로 기울여 한 발을 내딛는 mid-action이 성립하고, 리어카 측면·양쪽 손잡이·포댓자루 두 개가 모두 각 1/3 미만으로 읽히는 와이드 구도와 야간 저조도 로케이션 일치가 우수하다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "체크셔츠 청년이 손잡이를 전혀 쥐지 않고 따로 떨어져 있어 핵심 동작이 무너졌고, 포댓자루가 두 개가 아닌 다수이며 전경에 허용되지 않은 제3의 인물이 크게 들어와 하드 위반이다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 청년 모두 화면 오른쪽(숲 깊은 쪽 길)을 향해 시선과 몸을 두고 있고, 리어카는 카메라 축을 가로질러 우측으로 전진한다. 청년 #2(체크셔츠)의 두 손은 리어카 손잡이 두 개를 실제로 쥐고 있으며 팔이 손잡이 방향과 일치한다. 청년 #1(누더기·두건)은 그 앞쪽에서 같은 방향으로 몸을 기울인다. 프레임 좌측 가장자리에는 관찰자(김지후 추정)의 팔·다리 일부만 잘려 들어와 있어 은신 관찰자 옆 시점이라는 설정과 부합한다.",
      "built_space": "리어카는 측면이 보이며 바퀴 2개, 짐칸, 손잡이 두 개가 모두 한 세트로 정확히 렌더링됨. 손잡이는 뒤쪽(좌측)으로 뻗어 뒤에서 미는 사람의 손에 닿는다. 장소는 레퍼런스와 같은 해송림 좁은 길, 좌측 열 지어 선 이끼 낀 돌들(선돌 열)까지 일치.",
      "entities": "청년 #2는 붉은 체크셔츠·데님·부츠로 레퍼런스와 일치. 청년 #1은 회색 두건과 누더기 상의로 레퍼런스 의상과 일치(얼굴은 어둡게 처리). 포댓자루 두 개는 마대 질감에 어두운 붉은 얼룩, 상단 끈 묶음까지 프롭 레퍼런스와 부합하며 각각 화면 1/3 미만 차지. 좌측 잘린 인물은 얼굴 없는 부분 프레임이라 추가 인물 신설로 보긴 어려우나 신원 확인은 불가.",
      "hard_violations": [],
      "physics": "두 청년 모두 한쪽 발이 지면에 확실히 닿아 있고 앞으로 기운 상반신·뻗은 뒷다리로 밀고 나가는 하중이 읽힌다. 리어카는 두 바퀴가 땅에 닿아 있고 손잡이는 손에 지지됨. 공중에 뜬 물체 없음."
     },
     {
      "label": "B",
      "direction": "우측 청년 #1은 화면 오른쪽을 보며 리어카 손잡이 하나를 쥐고 앞으로 기울어 있다. 그러나 두 번째 청년(체크셔츠)은 리어카에서 멀리 떨어져 뒤쪽 길 위에 홀로 서서 카메라 쪽(좌측 전방)을 향해 몸을 굽히고 있어 손잡이를 전혀 쥐고 있지 않다 — 샷 텍스트의 '두 청년이 손잡이를 쥔 채'와 어긋난다. 좌측 전경에는 뼈 목걸이를 한 여성 인물의 목·가슴이 크게 들어와 있다.",
      "built_space": "리어카는 프레임 우측 가장자리에 잘려 짐칸 일부와 바퀴 하나만 보이고, 손잡이는 한쪽만 확인됨(양쪽 손잡이 모두 보이라는 지시 미충족). 나무 수레 형태로 현대식 리어카 특징(공기 타이어·철제 프레임)이 약함. 장소의 선돌 열과 해송은 레퍼런스와 부합.",
      "entities": "청년 #2 체크셔츠·데님·부츠 일치, 청년 #1 두건·누더기 일치. 포댓자루는 4개 이상 보이며 '두 개'라는 지시와 불일치, 마대 질감·붉은 얼룩은 프롭과 유사. 좌측 전경 여성 인물(김지후로 보이는 목걸이 착용자)이 얼굴 없이 크게 들어와 있으나 인물 목록에는 두 청년만 허용되어 있다.",
      "hard_violations": [
       "샷 텍스트가 지정한 인물 외 제3의 인물(뼈 목걸이 착용 여성)이 전경에 상반신 규모로 등장",
       "포댓자루가 두 개가 아니라 네 개 이상으로 렌더링됨"
      ],
      "physics": "두 인물 모두 한 발이 땅에 닿아 있고 지지는 있으나, 체크셔츠 청년은 손에 아무것도 없이 허리만 굽힌 어정쩡한 자세로 리어카를 미는 동작이 성립하지 않는다. 수레는 바퀴로 지지됨."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 청년 모두 시선이 화면 좌측(카메라 쪽/전경의 관찰자 방향)을 향해 있어, 프롬프트가 지정한 '숲 깊은 곳(화면 우측 안쪽) 루트로 전진'하는 방향과 어긋난다. 리어카는 화면 우측에 있고 손잡이는 좌측으로 뻗어 있어 이동 방향이 좌측(즉 화면 밖 왼쪽)으로 읽히며, 배경 루트도 좌측 안쪽으로 이어진다. 청년 #1은 오른손으로 손잡이를 쥐고 있지만 몸은 전진이 아니라 옆/뒤를 돌아본다. 청년 #2(체크셔츠)는 손잡이를 전혀 잡지 않고 빈손으로 웅크린 채 좌측을 본다.",
      "built_space": "리어카는 목재 짐칸에 나무 손잡이 두 개, 바퀴는 우측 하단에 살짝 보이는 금속 스포크 휠. 짐칸 위 포대는 두 개가 아니라 최소 3~4개가 겹쳐 보인다(우측 프레임에 큰 자루 3개 이상). 배경 좌측에 로케이션 사진의 이끼 낀 선돌 열이 재현되어 장소 락은 대체로 지켜짐. 전경 좌측에 나무 뒤 인물(목걸이 착용 여성)의 상반신이 크게 배치되어 있는데, 샷 텍스트가 지정한 등장인물 목록(마을 청년 #1, #2)에 없는 인물이 실체로 프레임 안에 들어와 있다.",
      "entities": "청년 #1: 두건과 누더기 한복 — 레퍼런스 일치. 청년 #2: 붉은 체크셔츠·데님·워크부츠 — 레퍼런스 일치. 포댓자루: 마대에 붉은 혈흔 — 프롭 레퍼런스와 일치하나 개수가 지정된 2개를 초과. 전경 여성: 프롬프트 PEOPLE 목록에 없는 제3의 인물(김지후로 추정되나 '숨은 관찰자' 곁 카메라 위치만 지정했을 뿐 등장 허용 대상 아님).",
      "hard_violations": [
       "샷 텍스트가 허용하지 않은 제3의 인물(전경 좌측 여성 상반신)이 프레임 내 실체로 등장 — PEOPLE 락 위반",
       "포댓자루가 지정된 2개가 아니라 3~4개로 복제되어 실림",
       "청년 #2가 리어카 손잡이를 전혀 쥐고 있지 않아 '두 청년이 손잡이를 쥔 채' 스테이징 불성립"
      ],
      "physics": "청년 #1은 양발이 지면에 있고 상체를 앞으로 기울여 손잡이를 잡은 자세로 지지는 성립. 청년 #2는 앞발/뒷발이 지면에 있으나 팔이 허공에 떠 있고 잡은 것이 없어 '리어카를 끄는' 힘의 전달이 없다. 리어카는 바퀴와 지면 접지가 보이며 자루들은 짐칸 위에 놓여 있어 지지는 있음. 자루 내부의 꿈틀거림(불룩함)은 거의 읽히지 않음."
     },
     {
      "label": "B",
      "direction": "두 청년 모두 시선과 몸통이 화면 우측 깊은 곳(숲 안쪽 루트)을 향하고, 리어카는 카메라 축을 가로질러 좌→우 측면으로 이동한다. 앞선 청년(체크셔츠, 화면 중우측)이 손잡이를 두 손으로 쥐고 앞으로 기울였고, 두 번째 청년(두건, 중앙)이 그 뒤/옆에서 밀며 같은 방향을 본다. 배경 루트도 우측 안쪽으로 이어져 FRAME LAYOUT과 일치.",
      "built_space": "리어카는 금속 프레임 짐칸에 큰 스포크 바퀴 하나가 명확히 보이고(반대편 바퀴는 가려짐), 손잡이 두 개가 우측으로 뻗어 청년의 손에 들어간다 — 실제 리어카 구조로 옳음. 짐칸 위 마대자루는 정확히 2개, 각각 화면의 1/3 미만을 차지. 배경 좌측에 로케이션 레퍼런스의 선돌 열과 해송 숲, 솔잎 깔린 흙길이 그대로 재현되어 장소 락 일치. 좌측 가장자리에 관찰자로 보이는 인물의 팔/다리 일부가 잘려 들어와 있으나 얼굴 없는 부분 프레임 진입 수준.",
      "entities": "앞 청년: 붉은 체크셔츠, 데님, 워크부츠, 짧은 검은 머리 — 청년 #2 레퍼런스 일치. 뒤 청년: 두건과 누더기 한복 — 청년 #1 레퍼런스 일치(단 FRAME LAYOUT의 좌우 배치는 서로 바뀐 편). 마대자루 2개: 거친 삼베, 끈으로 묶인 목, 붉은 혈흔 — 프롭 레퍼런스와 일치. 야간·저광량 렌더링으로 어둠 락 충족.",
      "hard_violations": [],
      "physics": "앞 청년은 뒷발이 지면을 밀고 앞발이 착지해 체중이 앞으로 실린 상태로 손잡이를 쥐고 있어 추진력이 성립. 뒤 청년도 양발 접지, 상체 전경사. 리어카는 바퀴가 지면에 닿아 있고 손잡이 하중이 사람 손으로 전달됨. 자루는 짐칸 안에 놓여 접지, 윤곽에 불룩한 볼륨이 있어 내부 움직임이 어느 정도 읽힘. 공중에 뜬 물체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "등장 불가 인물(전경 여성)이 실체로 크게 들어오고 포댓자루가 2개가 아닌 3개 이상으로 복제되었으며, 청년 #2가 손잡이를 쥐지 않아 '두 청년이 손잡이를 쥔 채 전진' 스테이징과 시선 방향(숲 안쪽) 모두 어긋난다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "와이드 측면 패닝, 손잡이를 쥔 두 청년의 전경사 자세, 정확히 2개의 혈흔 마대, 우측 숲 안쪽 진행 방향, 로케이션 선돌·해송길 일치까지 지정 스테이징을 충실히 실현했다(좌우 배치가 레이아웃과 살짝 반대인 점만 감점)."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "등장 불가 인물(전경 여성)이 실체로 크게 들어오고 포댓자루가 2개가 아닌 3개 이상으로 복제되었으며, 청년 #2가 손잡이를 쥐지 않아 '두 청년이 손잡이를 쥔 채 전진' 스테이징과 시선 방향(숲 안쪽) 모두 어긋난다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "와이드 측면 패닝, 손잡이를 쥔 두 청년의 전경사 자세, 정확히 2개의 혈흔 마대, 우측 숲 안쪽 진행 방향, 로케이션 선돌·해송길 일치까지 지정 스테이징을 충실히 실현했다(좌우 배치가 레이아웃과 살짝 반대인 점만 감점)."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "두 청년 모두 시선이 화면 좌측(카메라 쪽/전경의 관찰자 방향)을 향해 있어, 프롬프트가 지정한 '숲 깊은 곳(화면 우측 안쪽) 루트로 전진'하는 방향과 어긋난다. 리어카는 화면 우측에 있고 손잡이는 좌측으로 뻗어 있어 이동 방향이 좌측(즉 화면 밖 왼쪽)으로 읽히며, 배경 루트도 좌측 안쪽으로 이어진다. 청년 #1은 오른손으로 손잡이를 쥐고 있지만 몸은 전진이 아니라 옆/뒤를 돌아본다. 청년 #2(체크셔츠)는 손잡이를 전혀 잡지 않고 빈손으로 웅크린 채 좌측을 본다.",
      "built_space": "리어카는 목재 짐칸에 나무 손잡이 두 개, 바퀴는 우측 하단에 살짝 보이는 금속 스포크 휠. 짐칸 위 포대는 두 개가 아니라 최소 3~4개가 겹쳐 보인다(우측 프레임에 큰 자루 3개 이상). 배경 좌측에 로케이션 사진의 이끼 낀 선돌 열이 재현되어 장소 락은 대체로 지켜짐. 전경 좌측에 나무 뒤 인물(목걸이 착용 여성)의 상반신이 크게 배치되어 있는데, 샷 텍스트가 지정한 등장인물 목록(마을 청년 #1, #2)에 없는 인물이 실체로 프레임 안에 들어와 있다.",
      "entities": "청년 #1: 두건과 누더기 한복 — 레퍼런스 일치. 청년 #2: 붉은 체크셔츠·데님·워크부츠 — 레퍼런스 일치. 포댓자루: 마대에 붉은 혈흔 — 프롭 레퍼런스와 일치하나 개수가 지정된 2개를 초과. 전경 여성: 프롬프트 PEOPLE 목록에 없는 제3의 인물(김지후로 추정되나 '숨은 관찰자' 곁 카메라 위치만 지정했을 뿐 등장 허용 대상 아님).",
      "hard_violations": [
       "샷 텍스트가 허용하지 않은 제3의 인물(전경 좌측 여성 상반신)이 프레임 내 실체로 등장 — PEOPLE 락 위반",
       "포댓자루가 지정된 2개가 아니라 3~4개로 복제되어 실림",
       "청년 #2가 리어카 손잡이를 전혀 쥐고 있지 않아 '두 청년이 손잡이를 쥔 채' 스테이징 불성립"
      ],
      "physics": "청년 #1은 양발이 지면에 있고 상체를 앞으로 기울여 손잡이를 잡은 자세로 지지는 성립. 청년 #2는 앞발/뒷발이 지면에 있으나 팔이 허공에 떠 있고 잡은 것이 없어 '리어카를 끄는' 힘의 전달이 없다. 리어카는 바퀴와 지면 접지가 보이며 자루들은 짐칸 위에 놓여 있어 지지는 있음. 자루 내부의 꿈틀거림(불룩함)은 거의 읽히지 않음."
     },
     {
      "label": "A",
      "direction": "두 청년 모두 시선과 몸통이 화면 우측 깊은 곳(숲 안쪽 루트)을 향하고, 리어카는 카메라 축을 가로질러 좌→우 측면으로 이동한다. 앞선 청년(체크셔츠, 화면 중우측)이 손잡이를 두 손으로 쥐고 앞으로 기울였고, 두 번째 청년(두건, 중앙)이 그 뒤/옆에서 밀며 같은 방향을 본다. 배경 루트도 우측 안쪽으로 이어져 FRAME LAYOUT과 일치.",
      "built_space": "리어카는 금속 프레임 짐칸에 큰 스포크 바퀴 하나가 명확히 보이고(반대편 바퀴는 가려짐), 손잡이 두 개가 우측으로 뻗어 청년의 손에 들어간다 — 실제 리어카 구조로 옳음. 짐칸 위 마대자루는 정확히 2개, 각각 화면의 1/3 미만을 차지. 배경 좌측에 로케이션 레퍼런스의 선돌 열과 해송 숲, 솔잎 깔린 흙길이 그대로 재현되어 장소 락 일치. 좌측 가장자리에 관찰자로 보이는 인물의 팔/다리 일부가 잘려 들어와 있으나 얼굴 없는 부분 프레임 진입 수준.",
      "entities": "앞 청년: 붉은 체크셔츠, 데님, 워크부츠, 짧은 검은 머리 — 청년 #2 레퍼런스 일치. 뒤 청년: 두건과 누더기 한복 — 청년 #1 레퍼런스 일치(단 FRAME LAYOUT의 좌우 배치는 서로 바뀐 편). 마대자루 2개: 거친 삼베, 끈으로 묶인 목, 붉은 혈흔 — 프롭 레퍼런스와 일치. 야간·저광량 렌더링으로 어둠 락 충족.",
      "hard_violations": [],
      "physics": "앞 청년은 뒷발이 지면을 밀고 앞발이 착지해 체중이 앞으로 실린 상태로 손잡이를 쥐고 있어 추진력이 성립. 뒤 청년도 양발 접지, 상체 전경사. 리어카는 바퀴가 지면에 닿아 있고 손잡이 하중이 사람 손으로 전달됨. 자루는 짐칸 안에 놓여 접지, 윤곽에 불룩한 볼륨이 있어 내부 움직임이 어느 정도 읽힘. 공중에 뜬 물체 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 6
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "readings": [
   {
    "label": "A",
    "direction": "두 청년 모두 화면 오른쪽(숲 깊은 쪽 길)을 향해 시선과 몸을 두고 있고, 리어카는 카메라 축을 가로질러 우측으로 전진한다. 청년 #2(체크셔츠)의 두 손은 리어카 손잡이 두 개를 실제로 쥐고 있으며 팔이 손잡이 방향과 일치한다. 청년 #1(누더기·두건)은 그 앞쪽에서 같은 방향으로 몸을 기울인다. 프레임 좌측 가장자리에는 관찰자(김지후 추정)의 팔·다리 일부만 잘려 들어와 있어 은신 관찰자 옆 시점이라는 설정과 부합한다.",
    "built_space": "리어카는 측면이 보이며 바퀴 2개, 짐칸, 손잡이 두 개가 모두 한 세트로 정확히 렌더링됨. 손잡이는 뒤쪽(좌측)으로 뻗어 뒤에서 미는 사람의 손에 닿는다. 장소는 레퍼런스와 같은 해송림 좁은 길, 좌측 열 지어 선 이끼 낀 돌들(선돌 열)까지 일치.",
    "entities": "청년 #2는 붉은 체크셔츠·데님·부츠로 레퍼런스와 일치. 청년 #1은 회색 두건과 누더기 상의로 레퍼런스 의상과 일치(얼굴은 어둡게 처리). 포댓자루 두 개는 마대 질감에 어두운 붉은 얼룩, 상단 끈 묶음까지 프롭 레퍼런스와 부합하며 각각 화면 1/3 미만 차지. 좌측 잘린 인물은 얼굴 없는 부분 프레임이라 추가 인물 신설로 보긴 어려우나 신원 확인은 불가.",
    "hard_violations": [],
    "physics": "두 청년 모두 한쪽 발이 지면에 확실히 닿아 있고 앞으로 기운 상반신·뻗은 뒷다리로 밀고 나가는 하중이 읽힌다. 리어카는 두 바퀴가 땅에 닿아 있고 손잡이는 손에 지지됨. 공중에 뜬 물체 없음."
   },
   {
    "label": "B",
    "direction": "우측 청년 #1은 화면 오른쪽을 보며 리어카 손잡이 하나를 쥐고 앞으로 기울어 있다. 그러나 두 번째 청년(체크셔츠)은 리어카에서 멀리 떨어져 뒤쪽 길 위에 홀로 서서 카메라 쪽(좌측 전방)을 향해 몸을 굽히고 있어 손잡이를 전혀 쥐고 있지 않다 — 샷 텍스트의 '두 청년이 손잡이를 쥔 채'와 어긋난다. 좌측 전경에는 뼈 목걸이를 한 여성 인물의 목·가슴이 크게 들어와 있다.",
    "built_space": "리어카는 프레임 우측 가장자리에 잘려 짐칸 일부와 바퀴 하나만 보이고, 손잡이는 한쪽만 확인됨(양쪽 손잡이 모두 보이라는 지시 미충족). 나무 수레 형태로 현대식 리어카 특징(공기 타이어·철제 프레임)이 약함. 장소의 선돌 열과 해송은 레퍼런스와 부합.",
    "entities": "청년 #2 체크셔츠·데님·부츠 일치, 청년 #1 두건·누더기 일치. 포댓자루는 4개 이상 보이며 '두 개'라는 지시와 불일치, 마대 질감·붉은 얼룩은 프롭과 유사. 좌측 전경 여성 인물(김지후로 보이는 목걸이 착용자)이 얼굴 없이 크게 들어와 있으나 인물 목록에는 두 청년만 허용되어 있다.",
    "hard_violations": [
     "샷 텍스트가 지정한 인물 외 제3의 인물(뼈 목걸이 착용 여성)이 전경에 상반신 규모로 등장",
     "포댓자루가 두 개가 아니라 네 개 이상으로 렌더링됨"
    ],
    "physics": "두 인물 모두 한 발이 땅에 닿아 있고 지지는 있으나, 체크셔츠 청년은 손에 아무것도 없이 허리만 굽힌 어정쩡한 자세로 리어카를 미는 동작이 성립하지 않는다. 수레는 바퀴로 지지됨."
   }
  ],
  "totals": {
   "A": 16,
   "B": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "두 청년이 실제로 리어카 손잡이를 쥐고 앞으로 기울여 한 발을 내딛는 mid-action이 성립하고, 리어카 측면·양쪽 손잡이·포댓자루 두 개가 모두 각 1/3 미만으로 읽히는 와이드 구도와 야간 저조도 로케이션 일치가 우수하다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "체크셔츠 청년이 손잡이를 전혀 쥐지 않고 따로 떨어져 있어 핵심 동작이 무너졌고, 포댓자루가 두 개가 아닌 다수이며 전경에 허용되지 않은 제3의 인물이 크게 들어와 하드 위반이다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송_숲길_cf7470.png"
   },
   {
    "label": "CHARACTER REFERENCE — 마을 청년 #1: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:976931>"
   },
   {
    "label": "CHARACTER REFERENCE — 마을 청년 #2: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1020350>"
   },
   {
    "label": "PROP REFERENCE — 피 묻은 포댓자루: the exact object appearing in this shot; match its look, material and wear exactly.",
    "path": "<bytes:1566892>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "화면 왼쪽 가장자리에 프롬프트가 허용하지 않은 제3의 인물(하반신 일부)이 등장함.",
     "fix_en": "Remove the partial figure on the far left edge of the frame, replacing that area with the dark dirt path and background pine trees. Preserve the two youths and their positions, their clothing, the cart with sacks, the set, the light, and the framing.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "리어카의 양쪽 손잡이가 보여야 한다는 지시와 달리, 하나의 얇은 외가닥 막대기로 표현됨.",
     "fix_en": "Replace the single thin tow pole with two parallel metal handcart handles extending forward from the cart, showing the youths' hands gripping them. Preserve the two youths and their positions, their clothing, the cart body and sacks, the set, the light, and the framing.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 6,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the partial figure on the far left edge of the frame, replacing that area with the dark dirt path and background pine trees. Preserve the two youths and their positions, their clothing, the cart with sacks, the set, the light, and the framing.\n- Replace the single thin tow pole with two parallel metal handcart handles extending forward from the cart, showing the youths' hands gripping them. Preserve the two youths and their positions, their clothing, the cart body and sacks, the set, the light, and the framing.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 청년 모두 화면 오른쪽(숲 안쪽으로 이어지는 길)을 향해 시선과 몸을 기울이고 있어 레이아웃 스케치의 화살표 방향과 일치. 선두 청년(체크셔츠, 청년#2)은 프레임 중앙 우측에서 손잡이를 잡고 오른쪽으로 전진, 두 번째 청년(두건, 청년#1)은 중앙에서 리어카 옆/뒤를 밀며 같은 방향을 봄. 리어카 손잡이 두 개가 오른쪽으로 뻗어 청년#2의 양손에 닿아 있음. 화면 좌측 가장자리에 사람 다리/팔 일부(제3의 인물)가 잘려 들어와 있는데, 이는 스케치의 세 번째 인물(지후)에 해당하는 위치로 SHOT TEXT는 청년 둘만 언급하나 CARRIED STATE 절이 지후의 동행을 전제함.",
      "built_space": "리어카는 한 대, 바퀴 두 개(근경 큰 바퀴 하나 명확, 반대쪽 바퀴 그림자로 암시), 짐칸 목재 측판과 금속 프레임이 카메라 축을 가로질러 측면으로 보임. 손잡이 두 개가 오른쪽으로 뻗어 청년#2 양손에 정확히 잡힘. 포댓자루 두 개가 짐칸에 실려 있고 각각 화면 폭의 1/3을 넘지 않음. 배경의 돌 미륵/장승 열, 해송, 오른쪽의 굵은 나무 줄기, 하늘 틈새 모두 SHOT BACKGROUND와 동일한 원근·위치 유지.",
      "entities": "청년#2: 붉은 체크 셔츠, 짙은 청바지, 짧은 검은 머리, 마른 얼굴 — 레퍼런스와 일치. 청년#1: 머리에 두건, 누더기 갈색 천옷 — 레퍼런스 의상과 일치(다만 조선시대 복장이 현대 청년과 병치되는 것은 레퍼런스가 그렇게 지시). 포댓자루: 삼베 질감, 상단 끈 묶음, 붉은 혈흔 — PROP 레퍼런스와 일치. 좌측 가장자리 인물은 흰 상의·운동화 일부만 보임.",
      "hard_violations": [],
      "physics": "두 청년 모두 한 발이 지면에 닿아 있고 상체를 앞으로 기울인 추진 자세 — 밀기 동작으로 설명 가능. 리어카는 바퀴로 지면에 지지되고 손잡이는 청년#2의 두 손이 쥐고 있음. 포댓자루는 짐칸 바닥에 얹혀 있고 서로 기대어 있으며, 표면에 불룩한 융기가 보임. 공중에 떠 있는 물체 없음."
     },
     {
      "label": "B",
      "direction": "A와 동일하게 두 청년이 화면 오른쪽 숲길 방향으로 몸을 기울여 전진. 시선도 오른쪽. 그러나 리어카 손잡이 프레임이 왼쪽으로 길게 뻗은 여분의 막대 두 개로 렌더되어, 청년#2가 쥔 오른쪽 손잡이와 별개로 왼쪽에도 손잡이 형상이 생겼고 그 끝은 아무도 잡지 않은 채 허공에 떠 있음. 좌측 가장자리의 제3 인물은 삭제됨.",
      "built_space": "리어카 한 대이나 손잡이 구조가 이중으로 보임: 짐칸 오른쪽에서 청년#2에게 뻗은 정상 손잡이 한 쌍과, 짐칸 왼쪽으로 뻗어 나온 또 한 쌍의 긴 막대. 실제 리어카에는 한쪽에만 손잡이가 있어 이는 조종부 중복에 해당. 짐칸·바퀴·측판은 A와 동일. 배경(장승 열, 해송, 우측 큰 줄기)은 SHOT BACKGROUND와 일치하나 전체 톤이 A보다 밝고 평평해 '전화 불빛을 끈 뒤 매우 낮은 야간 가시도' 조건에서 다소 벗어남.",
      "entities": "인물 두 명의 의상·얼굴·머리는 A와 동일(청년#2 붉은 체크셔츠, 청년#1 두건과 누더기). 포댓자루 두 개, 혈흔 삼베 — 레퍼런스 일치. 좌측 인물(지후 추정)이 사라져 CARRIED STATE의 '지후가 수레를 따라간다'는 이웃 샷 연속성이 프레임에서 제거됨.",
      "hard_violations": [
       "리어카의 왼쪽으로 뻗은 여분의 손잡이 한 쌍 — 아무 손도 잡지 않은 채 허공에 떠 있는 중복 조종부"
      ],
      "physics": "인물의 발 접지와 전방 기울임은 A와 동일하게 성립. 그러나 짐칸 왼쪽으로 튀어나온 긴 손잡이 막대는 지면 지지도, 쥔 손도 없이 공중에 떠 있고 짐칸과의 접합부도 불명확함."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "손잡이 두 개가 청년#2의 양손에 정확히 잡히고 리어카 구조가 하나로 읽히며, 어둠 농도도 '폰 불빛을 끈 뒤' 야간 가시도 지시에 부합 — 좌측 가장자리 인물 조각과 카트가 다소 오른쪽으로 몰린 점만 약점."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "카트를 좌측으로 옮겨 배치를 넓힌 대신 아무도 쥐지 않은 여분의 손잡이 한 쌍이 허공에 생겼고, 좌측 인물 삭제와 전체적으로 밝아진 톤까지 계약이 요구하지 않은 손실이라 수리 비용이 이득을 넘어섬."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "손잡이 두 개가 청년#2의 양손에 정확히 잡히고 리어카 구조가 하나로 읽히며, 어둠 농도도 '폰 불빛을 끈 뒤' 야간 가시도 지시에 부합 — 좌측 가장자리 인물 조각과 카트가 다소 오른쪽으로 몰린 점만 약점."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "카트를 좌측으로 옮겨 배치를 넓힌 대신 아무도 쥐지 않은 여분의 손잡이 한 쌍이 허공에 생겼고, 좌측 인물 삭제와 전체적으로 밝아진 톤까지 계약이 요구하지 않은 손실이라 수리 비용이 이득을 넘어섬."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 청년 모두 화면 오른쪽(숲 안쪽으로 이어지는 길)을 향해 시선과 몸을 기울이고 있어 레이아웃 스케치의 화살표 방향과 일치. 선두 청년(체크셔츠, 청년#2)은 프레임 중앙 우측에서 손잡이를 잡고 오른쪽으로 전진, 두 번째 청년(두건, 청년#1)은 중앙에서 리어카 옆/뒤를 밀며 같은 방향을 봄. 리어카 손잡이 두 개가 오른쪽으로 뻗어 청년#2의 양손에 닿아 있음. 화면 좌측 가장자리에 사람 다리/팔 일부(제3의 인물)가 잘려 들어와 있는데, 이는 스케치의 세 번째 인물(지후)에 해당하는 위치로 SHOT TEXT는 청년 둘만 언급하나 CARRIED STATE 절이 지후의 동행을 전제함.",
      "built_space": "리어카는 한 대, 바퀴 두 개(근경 큰 바퀴 하나 명확, 반대쪽 바퀴 그림자로 암시), 짐칸 목재 측판과 금속 프레임이 카메라 축을 가로질러 측면으로 보임. 손잡이 두 개가 오른쪽으로 뻗어 청년#2 양손에 정확히 잡힘. 포댓자루 두 개가 짐칸에 실려 있고 각각 화면 폭의 1/3을 넘지 않음. 배경의 돌 미륵/장승 열, 해송, 오른쪽의 굵은 나무 줄기, 하늘 틈새 모두 SHOT BACKGROUND와 동일한 원근·위치 유지.",
      "entities": "청년#2: 붉은 체크 셔츠, 짙은 청바지, 짧은 검은 머리, 마른 얼굴 — 레퍼런스와 일치. 청년#1: 머리에 두건, 누더기 갈색 천옷 — 레퍼런스 의상과 일치(다만 조선시대 복장이 현대 청년과 병치되는 것은 레퍼런스가 그렇게 지시). 포댓자루: 삼베 질감, 상단 끈 묶음, 붉은 혈흔 — PROP 레퍼런스와 일치. 좌측 가장자리 인물은 흰 상의·운동화 일부만 보임.",
      "hard_violations": [],
      "physics": "두 청년 모두 한 발이 지면에 닿아 있고 상체를 앞으로 기울인 추진 자세 — 밀기 동작으로 설명 가능. 리어카는 바퀴로 지면에 지지되고 손잡이는 청년#2의 두 손이 쥐고 있음. 포댓자루는 짐칸 바닥에 얹혀 있고 서로 기대어 있으며, 표면에 불룩한 융기가 보임. 공중에 떠 있는 물체 없음."
     },
     {
      "label": "B",
      "direction": "A와 동일하게 두 청년이 화면 오른쪽 숲길 방향으로 몸을 기울여 전진. 시선도 오른쪽. 그러나 리어카 손잡이 프레임이 왼쪽으로 길게 뻗은 여분의 막대 두 개로 렌더되어, 청년#2가 쥔 오른쪽 손잡이와 별개로 왼쪽에도 손잡이 형상이 생겼고 그 끝은 아무도 잡지 않은 채 허공에 떠 있음. 좌측 가장자리의 제3 인물은 삭제됨.",
      "built_space": "리어카 한 대이나 손잡이 구조가 이중으로 보임: 짐칸 오른쪽에서 청년#2에게 뻗은 정상 손잡이 한 쌍과, 짐칸 왼쪽으로 뻗어 나온 또 한 쌍의 긴 막대. 실제 리어카에는 한쪽에만 손잡이가 있어 이는 조종부 중복에 해당. 짐칸·바퀴·측판은 A와 동일. 배경(장승 열, 해송, 우측 큰 줄기)은 SHOT BACKGROUND와 일치하나 전체 톤이 A보다 밝고 평평해 '전화 불빛을 끈 뒤 매우 낮은 야간 가시도' 조건에서 다소 벗어남.",
      "entities": "인물 두 명의 의상·얼굴·머리는 A와 동일(청년#2 붉은 체크셔츠, 청년#1 두건과 누더기). 포댓자루 두 개, 혈흔 삼베 — 레퍼런스 일치. 좌측 인물(지후 추정)이 사라져 CARRIED STATE의 '지후가 수레를 따라간다'는 이웃 샷 연속성이 프레임에서 제거됨.",
      "hard_violations": [
       "리어카의 왼쪽으로 뻗은 여분의 손잡이 한 쌍 — 아무 손도 잡지 않은 채 허공에 떠 있는 중복 조종부"
      ],
      "physics": "인물의 발 접지와 전방 기울임은 A와 동일하게 성립. 그러나 짐칸 왼쪽으로 튀어나온 긴 손잡이 막대는 지면 지지도, 쥔 손도 없이 공중에 떠 있고 짐칸과의 접합부도 불명확함."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 청년 모두 화면 왼쪽을 향해 시선과 몸을 두고 리어카를 밀며 전진한다. 리어카 손잡이는 오른쪽 뒤로 뻗어 뒤쪽 청년(체크셔츠)의 두 손에 잡혀 있고, 진행 방향은 화면 좌측 숲길이다. 레이아웃 스케치는 인물들이 화면 우측(숲 깊은 곳)으로 향하도록 지시하는데, 두 후보 모두 좌향으로 반전되어 있어 이 점은 동일. 시선은 앞쪽 길 위에 정확히 놓여 있다.",
      "built_space": "리어카는 한 대, 바퀴는 좌우 각 1개(측면에서 큰 바퀴 1개 및 반대편 바퀴 일부)가 보이고 손잡이 두 개가 하나의 프레임에서 뒤로 뻗어 정상. 짐칸에 포댓자루 두 개가 실려 있으며 각각 화면의 1/3 미만을 차지. 배경 석상 열, 해송 줄기, 우측 큰 나무 등 SHOT BACKGROUND의 고정 요소가 그대로 유지된다. 청년 #1(두건·누더기)은 리어카 옆 중앙, 청년 #2(체크셔츠)는 손잡이 쪽 우측으로 배치.",
      "entities": "청년 #1: 머리에 천 두건, 갈색 누더기 한복 — 레퍼런스와 일치하는 한국인 청년. 청년 #2: 붉은 체크 셔츠, 청바지, 짧은 검은 머리 — 레퍼런스 일치. 프레임 안에 제3의 인물 없음. 포댓자루는 마대 재질에 붉은 혈흔이 있어 프롭 레퍼런스와 부합. 텍스트·워터마크 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 발이 지면에 닿아 있고 상체를 앞으로 기울인 추진 자세. 뒤쪽 청년은 두 손으로 손잡이를 확실히 쥐고 팔이 뻗어 밀고 있으며 뒷발이 지면을 딛고 있다. 리어카는 바퀴로 지면에 지지되고 자루는 짐칸 위에 얹혀 있다. 공중에 떠 있는 물체 없음."
     },
     {
      "label": "B",
      "direction": "A와 동일한 좌향 진행·시선. 다만 화면 왼쪽 가장자리에 잘린 제3의 인물(흰 소매 팔과 회색 운동화, 다리)이 프레임 안으로 걸어 들어오는 것이 보이며, 이 인물의 진행 방향도 좌측이다. 주 인물들의 조준·시선은 A와 같이 앞길에 놓임.",
      "built_space": "리어카·바퀴·손잡이 구성은 A와 동일하게 정상. 그러나 프레임이 약간 다르게 재구성되어 좌하단 배경이 밝게 들리고, 좌측 끝에 배경에 없던 인체 부분이 추가됨. 배경 석상과 해송은 유지.",
      "entities": "청년 #1, #2는 A와 동일하게 레퍼런스 일치. 문제는 좌측 가장자리의 추가 인물 — 흰 셔츠 소매와 회색 스니커즈로 청년 #1·#2 어느 쪽과도 복장이 일치하지 않으며, 프롬프트가 허용한 두 인물 외의 제3자가 프레임에 들어와 있다.",
      "hard_violations": [
       "프롬프트가 허용하지 않은 제3의 인물(좌측 가장자리의 팔·다리·운동화)이 추가됨 — 'never add a person the shot text does not show' 위반"
      ],
      "physics": "주 인물과 리어카는 A와 동일하게 지면에 지지되어 문제없음. 좌측 추가 인물은 다리·발이 지면에 닿아 있으나 상반신이 프레임 밖으로 잘려 정체 불명의 유령 같은 부분 신체로 남는다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "배경·리어카·두 자루·두 청년 배치와 앞으로 기울인 mid-action이 모두 충족되고 계약 외 인물이 없다; 스케치의 우향 진행이 좌향으로 반전된 점만 감점."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "A와 거의 동일하나 좌측 가장자리에 계약이 허용하지 않은 제3의 인물(흰 소매·회색 운동화)이 침입해 'never add a person' 조항을 정면 위반하고, 좌하단 노출도 들떠 손해만 남긴 변경이다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "배경·리어카·두 자루·두 청년 배치와 앞으로 기울인 mid-action이 모두 충족되고 계약 외 인물이 없다; 스케치의 우향 진행이 좌향으로 반전된 점만 감점."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "A와 거의 동일하나 좌측 가장자리에 계약이 허용하지 않은 제3의 인물(흰 소매·회색 운동화)이 침입해 'never add a person' 조항을 정면 위반하고, 좌하단 노출도 들떠 손해만 남긴 변경이다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "두 청년 모두 화면 왼쪽을 향해 시선과 몸을 두고 리어카를 밀며 전진한다. 리어카 손잡이는 오른쪽 뒤로 뻗어 뒤쪽 청년(체크셔츠)의 두 손에 잡혀 있고, 진행 방향은 화면 좌측 숲길이다. 레이아웃 스케치는 인물들이 화면 우측(숲 깊은 곳)으로 향하도록 지시하는데, 두 후보 모두 좌향으로 반전되어 있어 이 점은 동일. 시선은 앞쪽 길 위에 정확히 놓여 있다.",
      "built_space": "리어카는 한 대, 바퀴는 좌우 각 1개(측면에서 큰 바퀴 1개 및 반대편 바퀴 일부)가 보이고 손잡이 두 개가 하나의 프레임에서 뒤로 뻗어 정상. 짐칸에 포댓자루 두 개가 실려 있으며 각각 화면의 1/3 미만을 차지. 배경 석상 열, 해송 줄기, 우측 큰 나무 등 SHOT BACKGROUND의 고정 요소가 그대로 유지된다. 청년 #1(두건·누더기)은 리어카 옆 중앙, 청년 #2(체크셔츠)는 손잡이 쪽 우측으로 배치.",
      "entities": "청년 #1: 머리에 천 두건, 갈색 누더기 한복 — 레퍼런스와 일치하는 한국인 청년. 청년 #2: 붉은 체크 셔츠, 청바지, 짧은 검은 머리 — 레퍼런스 일치. 프레임 안에 제3의 인물 없음. 포댓자루는 마대 재질에 붉은 혈흔이 있어 프롭 레퍼런스와 부합. 텍스트·워터마크 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 발이 지면에 닿아 있고 상체를 앞으로 기울인 추진 자세. 뒤쪽 청년은 두 손으로 손잡이를 확실히 쥐고 팔이 뻗어 밀고 있으며 뒷발이 지면을 딛고 있다. 리어카는 바퀴로 지면에 지지되고 자루는 짐칸 위에 얹혀 있다. 공중에 떠 있는 물체 없음."
     },
     {
      "label": "A",
      "direction": "A와 동일한 좌향 진행·시선. 다만 화면 왼쪽 가장자리에 잘린 제3의 인물(흰 소매 팔과 회색 운동화, 다리)이 프레임 안으로 걸어 들어오는 것이 보이며, 이 인물의 진행 방향도 좌측이다. 주 인물들의 조준·시선은 A와 같이 앞길에 놓임.",
      "built_space": "리어카·바퀴·손잡이 구성은 A와 동일하게 정상. 그러나 프레임이 약간 다르게 재구성되어 좌하단 배경이 밝게 들리고, 좌측 끝에 배경에 없던 인체 부분이 추가됨. 배경 석상과 해송은 유지.",
      "entities": "청년 #1, #2는 A와 동일하게 레퍼런스 일치. 문제는 좌측 가장자리의 추가 인물 — 흰 셔츠 소매와 회색 스니커즈로 청년 #1·#2 어느 쪽과도 복장이 일치하지 않으며, 프롬프트가 허용한 두 인물 외의 제3자가 프레임에 들어와 있다.",
      "hard_violations": [
       "프롬프트가 허용하지 않은 제3의 인물(좌측 가장자리의 팔·다리·운동화)이 추가됨 — 'never add a person the shot text does not show' 위반"
      ],
      "physics": "주 인물과 리어카는 A와 동일하게 지면에 지지되어 문제없음. 좌측 추가 인물은 다리·발이 지면에 닿아 있으나 상반신이 프레임 밖으로 잘려 정체 불명의 유령 같은 부분 신체로 남는다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 12
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S29sh2__bgfirst_bg.png",
   "bg_asset_id": "8bb225ff-4300-4acc-8286-35008779b609",
   "bg_record_key": "S29sh2::bgfirst_bg",
   "chain_winner": true,
   "authority": "groupbg",
   "group_key": "해송 숲길",
   "groupbg_asset_id": "26370ef0-070a-470c-beac-42f57a83eef4"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S30sh7": {
  "input_fingerprint": "639171742c11f98b",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 숲속 깊은 곳으로 향하는 금월도 마을 청년들의 뒤를 쫓아 김여상과 김지후(강수리영)가 조심스럽게 몸을 반쯤 일으킨 채 앞으로 한 발을 내딛는 mid-action 찰나.\n\nLOCATION (lock): A progressively narrower route through dense black-pine woodland beyond the marked trail. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Begin the flow’s low backward track several paces ahead of 김여상 and 김지후(강수리영), holding the moment both are half-risen and committing one cautious step toward the departing youths. Place the followers across the lower half of the frame while the youths recede back-to-camera near the upper center, preserving a clear single pursuit axis into the forest.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김여상 beginning pursuit in the lower-left of the frame, foreground, moves toward departing village youths; 김지후(강수리영) beginning pursuit in the lower-right of the frame, foreground, moves toward departing village youths; departing village youth in the upper-center of the frame, background, moves toward deeper forest.\n- KEY BACKGROUND ELEMENTS: 숲속 이동 경로 (Leading deeper into the forest) — The route recedes from the followers at the bottom toward the youths at the upper center; used as Creates the central depth axis separating the followers from the youths; 포댓자루 (Being carried deeper into the forest) — Seen from behind while carried on a youth's shoulder; used as Remains visible on the departing group as evidence of what is being followed.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low nighttime visibility renders the distance between pursuers and youths in heavy contrast.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same narrow forest route, pine trunks, darkness, cart trail, and receding village youths. Exclude the earlier cart close emphasis; show the two followers rising and continuing after them.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The youths continue carrying the two moving, blood-leaking sacks deeper into the woods; Jihu still wears her necklace.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face); 김여상 (Korean man in his mid-40s, dark hair, ordinary mature face); 마을 청년 #1 (Korean young man, short dark hair, broad youthful face); 마을 청년 #2 (Korean young man, short dark hair, lean youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 6,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "김여상 의상 연속성은 틀리지만, 지정된 하단 좌우 배치와 상단 중앙의 두 청년·두 자루, 그리고 숲 안쪽으로 이어지는 실제 추적 방향을 유일하게 정확히 구현했다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "인물 수와 이전 장면 의상, 두 포댓자루는 가장 정확하지만 두 추적자가 청년이 아닌 카메라 쪽으로 걷는 방향 오류가 핵심 추적 행동을 훼손한다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "추적자들의 이동 방향이 청년들과 반대이고 배경에 허용되지 않은 추가 인물까지 있어 사용 불가능하며, 두 포댓자루도 명확히 충족하지 못한다.  ★위반: [claude-opus] 상단 중앙에 허용된 마을 청년 두 명 외의 추가 인물 한 명이 생성되었다. / [gpt] 배경에 마을 청년이 세 명 나타나 허용된 두 명 외의 추가 인물이 생겼다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S30sh4_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:813104>"
   },
   {
    "label": "CHARACTER REFERENCE — 김여상: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:990143>"
   },
   {
    "label": "CHARACTER REFERENCE — 마을 청년 #1: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:976931>"
   },
   {
    "label": "CHARACTER REFERENCE — 마을 청년 #2: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1020350>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "[Critical] 배경 좌측 인물이 마을 청년이 아닌 김지후의 단독 레퍼런스 복장(검은 모자, 어두운 재킷)과 긴 머리를 하고 있어 인물이 잘못 적용되었습니다.",
     "fix_en": "Change the left background person to Village Youth #1 wearing a tattered, ragged brown hanbok and cloth headwrap. Keep them carrying the sack. Preserve exactly the right background person, both foreground characters, their poses, the forest, and framing.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "[Critical] 좌측 근경 남성의 복장이 이전 샷 스틸(파란색/주황색 재킷과 녹색 모자)과 연속되지 않고 단독 레퍼런스의 초록색 우비로 렌더링되었습니다.",
     "fix_en": "Change the clothing of the lower-left foreground man to match the previous still: a bright blue and orange technical jacket and a green baseball cap. Keep his exact pose. Preserve exactly the woman on the right, background figures, forest path, and framing.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "[Minor] 우측 근경 여성의 뒷주머니에 꽂힌 휴대폰 위치가 이전 샷(왼쪽 주머니)과 달리 오른쪽 주머니로 바뀌었습니다.",
     "fix_en": "",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 6,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the left background person to Village Youth #1 wearing a tattered, ragged brown hanbok and cloth headwrap. Keep them carrying the sack. Preserve exactly the right background person, both foreground characters, their poses, the forest, and framing.\n- Change the clothing of the lower-left foreground man to match the previous still: a bright blue and orange technical jacket and a green baseball cap. Keep his exact pose. Preserve exactly the woman on the right, background figures, forest path, and framing.\n- \n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 낮은 위치에서 뒤로 트래킹하며 숲길 축을 정면으로 잡는다. 좌하단 김여상은 등을 보인 채 숲 안쪽(상단 중앙)으로 향하고, 우하단 김지후도 몸을 반쯤 일으켜 같은 축을 향한다. 두 추적자의 시선·진행 방향 모두 상단 중앙의 두 청년에게 정확히 꽂힌다. 청년 2명은 카메라에 등을 보이며 더 깊은 숲으로 멀어진다 — 단일 추격 축이 명확.",
      "built_space": "인공 구조물은 없음. 검은 소나무 줄기가 좌우로 도열해 좁아지는 수레길이 소실점으로 수렴, 이전 스틸의 낙엽 깔린 좁은 길·잡목 덤불과 재질이 일치. 배경 요소가 전경으로 과장 확대된 것 없음.",
      "entities": "김여상: 짙은 녹색 방수 재킷(캐릭터 레퍼런스의 녹색 고무 재킷 계열)이나 이전 스틸의 파란 낚시 재킷+다이와 캡과는 불일치 — 의상 연속성 이탈. 김지후: 긴 흑발, 이전 스틸의 케이블 니트+청바지+뒷주머니 휴대폰까지 일치. 청년 2명: 한 명은 붉은 체크셔츠(청년#2 일치), 다른 한 명은 어두운 옷의 실루엣으로 정체 불명확하나 현대복. 포댓자루 2개가 각각 어깨/손에 들려 뒤에서 보임 — 캐리 상태 충족.",
      "hard_violations": [],
      "physics": "김여상은 한 발이 지면에 닿고 다른 발이 앞으로 나가는 반쯤 일어선 전진 자세, 상체가 앞으로 기울어 무게 이동이 읽힌다. 김지후는 웅크렸다 일어서는 중간 자세로 다리가 프레임 하단 지면에 접지. 청년들은 정상 보행, 자루 무게로 어깨가 눌린다. 공중에 뜬 물체 없음."
     },
     {
      "label": "B",
      "direction": "좌측 김여상은 몸을 일으켰으나 얼굴과 시선이 카메라 쪽(좌측 전방)을 향해 있어 추격 대상인 청년들을 보지 않는다 — 시선 축 이탈. 우하단 김지후는 등을 보이며 숲 안쪽을 향해 맞음. 중앙의 청년(누더기 한복 차림)은 카메라를 정면으로 바라보며 서 있어 '멀어지는 뒷모습' 지시와 정반대. 우측 체크셔츠 청년만 등을 보이고 멀어진다.",
      "built_space": "동일한 소나무 길·소실점 구조로 장소 락은 유지. 다만 좌측 김여상이 원근상 지나치게 크게 전경을 차지해 상반신 미디엄에 가까운 크기로 확대됨.",
      "entities": "김여상: 이전 스틸과 동일한 파란/주황 낚시 재킷+다이와 캡으로 의상 연속성은 정확. 김지후: A와 동일. 중앙 인물은 1670 조선 시대 복장의 마을 청년 #1 레퍼런스를 그대로 현대 야간 숲에 배치 — 시대 혼입이며 자루도 들지 않은 채 정면 정지. 포댓자루는 우측 청년 1개만 보임(두 개여야 함).",
      "hard_violations": [
       "추격 대상 청년이 카메라를 정면으로 응시하며 정지 — 샷 텍스트가 명시한 '멀어지는 뒷모습' 스테이징을 정반대로 렌더",
       "김여상이 추격 방향이 아닌 카메라 쪽을 보며 전경으로 확대되어 단일 추격 축이 붕괴"
      ],
      "physics": "김여상은 두 발로 접지해 서 있으나 전진 스텝의 무게 이동이 약함. 중앙 청년은 두 발로 서서 완전 정지 — mid-action 아님. 우측 청년은 자루를 어깨에 메고 보행 중, 지지 정상. 공중 부양물 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "저각 백트래킹 와이드, 하단 좌·우 추적자와 상단 중앙 등돌린 청년 2명·포댓자루 2개까지 프레임 레이아웃과 단일 추격 축을 정확히 실현했고 반쯤 일어선 한 발 내딛기 순간도 살아 있다. 김여상 상의가 이전 스틸의 파란 낚시 재킷 대신 녹색 재킷인 점이 유일한 연속성 손실."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "김여상 의상 연속성은 회복했으나, 그 대가로 김여상을 전경에 확대해 카메라를 보게 만들고 중앙 청년을 정지·정면 응시의 조선시대 복장 인물로 바꿔 '멀어지는 뒷모습' 추격 축과 포댓자루 2개 조건을 모두 무너뜨렸다 — 고친 것보다 잃은 것이 훨씬 크다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "저각 백트래킹 와이드, 하단 좌·우 추적자와 상단 중앙 등돌린 청년 2명·포댓자루 2개까지 프레임 레이아웃과 단일 추격 축을 정확히 실현했고 반쯤 일어선 한 발 내딛기 순간도 살아 있다. 김여상 상의가 이전 스틸의 파란 낚시 재킷 대신 녹색 재킷인 점이 유일한 연속성 손실."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "김여상 의상 연속성은 회복했으나, 그 대가로 김여상을 전경에 확대해 카메라를 보게 만들고 중앙 청년을 정지·정면 응시의 조선시대 복장 인물로 바꿔 '멀어지는 뒷모습' 추격 축과 포댓자루 2개 조건을 모두 무너뜨렸다 — 고친 것보다 잃은 것이 훨씬 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 낮은 위치에서 뒤로 트래킹하며 숲길 축을 정면으로 잡는다. 좌하단 김여상은 등을 보인 채 숲 안쪽(상단 중앙)으로 향하고, 우하단 김지후도 몸을 반쯤 일으켜 같은 축을 향한다. 두 추적자의 시선·진행 방향 모두 상단 중앙의 두 청년에게 정확히 꽂힌다. 청년 2명은 카메라에 등을 보이며 더 깊은 숲으로 멀어진다 — 단일 추격 축이 명확.",
      "built_space": "인공 구조물은 없음. 검은 소나무 줄기가 좌우로 도열해 좁아지는 수레길이 소실점으로 수렴, 이전 스틸의 낙엽 깔린 좁은 길·잡목 덤불과 재질이 일치. 배경 요소가 전경으로 과장 확대된 것 없음.",
      "entities": "김여상: 짙은 녹색 방수 재킷(캐릭터 레퍼런스의 녹색 고무 재킷 계열)이나 이전 스틸의 파란 낚시 재킷+다이와 캡과는 불일치 — 의상 연속성 이탈. 김지후: 긴 흑발, 이전 스틸의 케이블 니트+청바지+뒷주머니 휴대폰까지 일치. 청년 2명: 한 명은 붉은 체크셔츠(청년#2 일치), 다른 한 명은 어두운 옷의 실루엣으로 정체 불명확하나 현대복. 포댓자루 2개가 각각 어깨/손에 들려 뒤에서 보임 — 캐리 상태 충족.",
      "hard_violations": [],
      "physics": "김여상은 한 발이 지면에 닿고 다른 발이 앞으로 나가는 반쯤 일어선 전진 자세, 상체가 앞으로 기울어 무게 이동이 읽힌다. 김지후는 웅크렸다 일어서는 중간 자세로 다리가 프레임 하단 지면에 접지. 청년들은 정상 보행, 자루 무게로 어깨가 눌린다. 공중에 뜬 물체 없음."
     },
     {
      "label": "B",
      "direction": "좌측 김여상은 몸을 일으켰으나 얼굴과 시선이 카메라 쪽(좌측 전방)을 향해 있어 추격 대상인 청년들을 보지 않는다 — 시선 축 이탈. 우하단 김지후는 등을 보이며 숲 안쪽을 향해 맞음. 중앙의 청년(누더기 한복 차림)은 카메라를 정면으로 바라보며 서 있어 '멀어지는 뒷모습' 지시와 정반대. 우측 체크셔츠 청년만 등을 보이고 멀어진다.",
      "built_space": "동일한 소나무 길·소실점 구조로 장소 락은 유지. 다만 좌측 김여상이 원근상 지나치게 크게 전경을 차지해 상반신 미디엄에 가까운 크기로 확대됨.",
      "entities": "김여상: 이전 스틸과 동일한 파란/주황 낚시 재킷+다이와 캡으로 의상 연속성은 정확. 김지후: A와 동일. 중앙 인물은 1670 조선 시대 복장의 마을 청년 #1 레퍼런스를 그대로 현대 야간 숲에 배치 — 시대 혼입이며 자루도 들지 않은 채 정면 정지. 포댓자루는 우측 청년 1개만 보임(두 개여야 함).",
      "hard_violations": [
       "추격 대상 청년이 카메라를 정면으로 응시하며 정지 — 샷 텍스트가 명시한 '멀어지는 뒷모습' 스테이징을 정반대로 렌더",
       "김여상이 추격 방향이 아닌 카메라 쪽을 보며 전경으로 확대되어 단일 추격 축이 붕괴"
      ],
      "physics": "김여상은 두 발로 접지해 서 있으나 전진 스텝의 무게 이동이 약함. 중앙 청년은 두 발로 서서 완전 정지 — mid-action 아님. 우측 청년은 자루를 어깨에 메고 보행 중, 지지 정상. 공중 부양물 없음."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "readings": [
     {
      "label": "A",
      "direction": "프레임 하단 좌측의 김여상은 몸을 일으킨 상태지만 얼굴과 시선이 카메라 쪽(왼쪽 앞)을 향하고 있어 숲 안쪽 청년들을 쫓는 시선축이 성립하지 않는다. 하단 우측 지후는 등을 보이며 숲 안쪽으로 향해 있어 방향은 맞다. 상단 중앙의 청년 두 명 중 오른쪽(체크셔츠, 포댓자루 어깨에 멘 청년#2)은 등을 보이고 숲 깊은 곳으로 이동 중이나, 왼쪽의 조선시대 누더기 차림 인물은 카메라를 정면으로 바라보며 서 있어 '떠나가는 청년들의 뒷모습'이라는 지시와 정면 충돌한다.",
      "built_space": "흑송림의 좁은 오솔길, 곧게 선 소나무 기둥 열, 낙엽 깔린 카트 트레일 — 이전 스틸의 장소/재질과 일관. 경로가 상단 중앙 소실점으로 수렴해 깊이축은 형성됨.",
      "entities": "김여상: 이전 스틸의 파란 낚시 재킷·다이와 로고 초록 캡·손가락 없는 장갑을 그대로 착용(이전 스틸 일치, 다만 캐릭터 레퍼런스의 초록 방수복과는 별개로 이전 스틸이 우선). 지후: 뒷모습, 케이블니트+청바지+뒷주머니 휴대폰으로 이전 스틸 일치, 다만 얼굴·목걸이는 보이지 않음. 청년#2: 붉은 체크셔츠·청바지·포댓자루 어깨 — 일치. 왼쪽 인물은 1670 조선 누더기 한복 차림의 청년#1 레퍼런스로, 현대 야간 추격 장면에 시대의상이 그대로 들어와 상황상 이물감이 크고 게다가 정면 응시. 포댓자루는 1개만 보임(계약은 2개).",
      "hard_violations": [
       "떠나가는 청년들 중 한 명이 카메라를 정면 응시하며 정지해 있어 '뒷모습으로 멀어지는' 스테이징을 위반(직접 카메라 응시 금지 조항 포함)"
      ],
      "physics": "모든 인물의 발이 지면에 닿아 있고 체중 이동이 보임. 김여상은 반쯤 일어선 자세로 한 발 내딛는 중, 지후도 구부린 상체로 전진 중 — 부양체 없음."
     },
     {
      "label": "B",
      "direction": "하단 좌측 김여상은 등을 보이며 숲 안쪽으로 한 발 내딛는 중 — 시선·진행 방향이 상단 중앙 청년들에게 정확히 향함. 하단 우측 지후도 몸을 반쯤 숙인 채 등을 보이고 같은 축으로 전진. 상단 중앙 두 청년 모두 완전한 뒷모습으로 숲 깊은 곳으로 멀어짐. 단일 추격축이 하단→상단 중앙으로 명확히 성립.",
      "built_space": "동일한 흑송림 좁은 경로, 소나무 기둥 열, 낙엽 트레일 — 이전 스틸과 재질·구조 일관. 경로가 좁아지며 상단 중앙으로 수렴.",
      "entities": "김여상: 캐릭터 레퍼런스의 초록 방수 재킷 착용(이전 스틸의 파란 낚시복과 불일치 — 이 지점이 손실). 지후: 케이블니트·청바지·뒷주머니 휴대폰·긴 흑발 뒷모습으로 이전 스틸 일치. 청년 2인: 오른쪽은 붉은 체크셔츠(청년#2) 일치, 왼쪽은 어두운 옷으로 시대의상 판별 불가하나 현대 야경에 무리 없음. 포댓자루 2개가 각각 어깨에 걸려 있어 '두 개의 자루' 계약 충족.",
      "hard_violations": [],
      "physics": "네 인물 모두 접지, 김여상은 뒷발 지면·앞발 내딛는 중간 자세, 지후는 상체를 낮춘 반쯤 일어선 전진 자세로 무게중심이 앞발에 실려 있음. 공중 부양 요소 없음."
     }
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "두 추격자가 모두 등을 보이며 상단 중앙의 뒷모습 청년들을 향하는 단일 추격축, 반쯤 일어선 한 발 내딛는 찰나, 포댓자루 2개까지 충족 — 김여상의 상의가 이전 스틸의 파란 낚시복 대신 캐릭터 레퍼런스의 초록 방수복으로 바뀐 점이 유일한 손실이나, 정면 응시·시대의상 혼입 같은 치명적 위반이 없어 우세."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "김여상 의상은 이전 스틸과 일치하지만, 김여상이 카메라 쪽을 보고 청년 한 명이 조선 누더기 차림으로 정면 응시한 채 멈춰 서 있어 '뒤를 쫓는' 추격축과 '뒷모습으로 멀어지는 청년들' 지시를 정면 위반하고, 포댓자루도 1개뿐이다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "두 추격자가 모두 등을 보이며 상단 중앙의 뒷모습 청년들을 향하는 단일 추격축, 반쯤 일어선 한 발 내딛는 찰나, 포댓자루 2개까지 충족 — 김여상의 상의가 이전 스틸의 파란 낚시복 대신 캐릭터 레퍼런스의 초록 방수복으로 바뀐 점이 유일한 손실이나, 정면 응시·시대의상 혼입 같은 치명적 위반이 없어 우세."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "김여상 의상은 이전 스틸과 일치하지만, 김여상이 카메라 쪽을 보고 청년 한 명이 조선 누더기 차림으로 정면 응시한 채 멈춰 서 있어 '뒤를 쫓는' 추격축과 '뒷모습으로 멀어지는 청년들' 지시를 정면 위반하고, 포댓자루도 1개뿐이다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "프레임 하단 좌측의 김여상은 몸을 일으킨 상태지만 얼굴과 시선이 카메라 쪽(왼쪽 앞)을 향하고 있어 숲 안쪽 청년들을 쫓는 시선축이 성립하지 않는다. 하단 우측 지후는 등을 보이며 숲 안쪽으로 향해 있어 방향은 맞다. 상단 중앙의 청년 두 명 중 오른쪽(체크셔츠, 포댓자루 어깨에 멘 청년#2)은 등을 보이고 숲 깊은 곳으로 이동 중이나, 왼쪽의 조선시대 누더기 차림 인물은 카메라를 정면으로 바라보며 서 있어 '떠나가는 청년들의 뒷모습'이라는 지시와 정면 충돌한다.",
      "built_space": "흑송림의 좁은 오솔길, 곧게 선 소나무 기둥 열, 낙엽 깔린 카트 트레일 — 이전 스틸의 장소/재질과 일관. 경로가 상단 중앙 소실점으로 수렴해 깊이축은 형성됨.",
      "entities": "김여상: 이전 스틸의 파란 낚시 재킷·다이와 로고 초록 캡·손가락 없는 장갑을 그대로 착용(이전 스틸 일치, 다만 캐릭터 레퍼런스의 초록 방수복과는 별개로 이전 스틸이 우선). 지후: 뒷모습, 케이블니트+청바지+뒷주머니 휴대폰으로 이전 스틸 일치, 다만 얼굴·목걸이는 보이지 않음. 청년#2: 붉은 체크셔츠·청바지·포댓자루 어깨 — 일치. 왼쪽 인물은 1670 조선 누더기 한복 차림의 청년#1 레퍼런스로, 현대 야간 추격 장면에 시대의상이 그대로 들어와 상황상 이물감이 크고 게다가 정면 응시. 포댓자루는 1개만 보임(계약은 2개).",
      "hard_violations": [
       "떠나가는 청년들 중 한 명이 카메라를 정면 응시하며 정지해 있어 '뒷모습으로 멀어지는' 스테이징을 위반(직접 카메라 응시 금지 조항 포함)"
      ],
      "physics": "모든 인물의 발이 지면에 닿아 있고 체중 이동이 보임. 김여상은 반쯤 일어선 자세로 한 발 내딛는 중, 지후도 구부린 상체로 전진 중 — 부양체 없음."
     },
     {
      "label": "A",
      "direction": "하단 좌측 김여상은 등을 보이며 숲 안쪽으로 한 발 내딛는 중 — 시선·진행 방향이 상단 중앙 청년들에게 정확히 향함. 하단 우측 지후도 몸을 반쯤 숙인 채 등을 보이고 같은 축으로 전진. 상단 중앙 두 청년 모두 완전한 뒷모습으로 숲 깊은 곳으로 멀어짐. 단일 추격축이 하단→상단 중앙으로 명확히 성립.",
      "built_space": "동일한 흑송림 좁은 경로, 소나무 기둥 열, 낙엽 트레일 — 이전 스틸과 재질·구조 일관. 경로가 좁아지며 상단 중앙으로 수렴.",
      "entities": "김여상: 캐릭터 레퍼런스의 초록 방수 재킷 착용(이전 스틸의 파란 낚시복과 불일치 — 이 지점이 손실). 지후: 케이블니트·청바지·뒷주머니 휴대폰·긴 흑발 뒷모습으로 이전 스틸 일치. 청년 2인: 오른쪽은 붉은 체크셔츠(청년#2) 일치, 왼쪽은 어두운 옷으로 시대의상 판별 불가하나 현대 야경에 무리 없음. 포댓자루 2개가 각각 어깨에 걸려 있어 '두 개의 자루' 계약 충족.",
      "hard_violations": [],
      "physics": "네 인물 모두 접지, 김여상은 뒷발 지면·앞발 내딛는 중간 자세, 지후는 상체를 낮춘 반쯤 일어선 전진 자세로 무게중심이 앞발에 실려 있음. 공중 부양 요소 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 7
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S30sh4"
  }
 },
 "S31sh3::bgfirst_bg": {
  "input_fingerprint": "0ceff6e4b8ea6afa",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 신경을 곤두세운 채 검은 염소 떼 사이 빈 공간으로 조심스레 한쪽 발을 내디딘 김여상과 김지후(강수리영)의 모습.\n\nLOCATION (lock): The open-sided goat pen deep in the forest, among tightly gathered black goats.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Dolly in from waist height through the established open gap between the goats, remaining about two paces ahead of 김여상 and 김지후(강수리영). Frame them in a cautious full-body two-shot as each lowers their center of gravity and places one foot into the gap, while black goats form irregular foreground and side barriers without blocking their route.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김여상 testing the route in the middle-left of the frame, midground, moves toward open gap between the goats; 김지후(강수리영) following beside him in the middle-right of the frame, midground, moves toward open gap between the goats; open gap between the goats in the lower-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 검은 염소 떼 (Dozens gathered inside the pen) — Goats flank both sides of the camera-facing gap, with partial bodies crossing the frame edges; used as Forms living foreground and lateral occlusion around the pair's narrow passage; 염소 떼 사이 빈 공간 (Temporarily clear after the village youths passed) — The opening runs from the lower center toward the approaching pair; used as Defines the precise route into which both characters are stepping.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low nighttime visibility inside the goat enclosure emphasizes dark silhouettes and restrained facial detail.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 신경을 곤두세운 채 검은 염소 떼 사이 빈 공간으로 조심스레 한쪽 발을 내디딘 김여상과 김지후(강수리영)의 모습.\n\nLOCATION (lock): The open-sided goat pen deep in the forest, among tightly gathered black goats.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Dolly in from waist height through the established open gap between the goats, remaining about two paces ahead of 김여상 and 김지후(강수리영). Frame them in a cautious full-body two-shot as each lowers their center of gravity and places one foot into the gap, while black goats form irregular foreground and side barriers without blocking their route.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김여상 testing the route in the middle-left of the frame, midground, moves toward open gap between the goats; 김지후(강수리영) following beside him in the middle-right of the frame, midground, moves toward open gap between the goats; open gap between the goats in the lower-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 검은 염소 떼 (Dozens gathered inside the pen) — Goats flank both sides of the camera-facing gap, with partial bodies crossing the frame edges; used as Forms living foreground and lateral occlusion around the pair's narrow passage; 염소 떼 사이 빈 공간 (Temporarily clear after the village youths passed) — The opening runs from the lower center toward the approaching pair; used as Defines the precise route into which both characters are stepping.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low nighttime visibility inside the goat enclosure emphasizes dark silhouettes and restrained facial detail.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S31sh3__bgfirst_bg.png",
  "asset_id": "3bb5c2ec-11db-4118-8872-a7d74b2f31aa",
  "input_asset_ids": [
   "1c72234a-103b-4af8-a82e-05b2b3882b25",
   "4240a7d4-8dee-4a7f-9e73-b218ab04ccc6"
  ]
 },
 "S31sh3": {
  "input_fingerprint": "b2521b1f701fc803",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 신경을 곤두세운 채 검은 염소 떼 사이 빈 공간으로 조심스레 한쪽 발을 내디딘 김여상과 김지후(강수리영)의 모습.\n\nLOCATION (lock): The open-sided goat pen deep in the forest, among tightly gathered black goats. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Dolly in from waist height through the established open gap between the goats, remaining about two paces ahead of 김여상 and 김지후(강수리영). Frame them in a cautious full-body two-shot as each lowers their center of gravity and places one foot into the gap, while black goats form irregular foreground and side barriers without blocking their route.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김여상 testing the route in the middle-left of the frame, midground, moves toward open gap between the goats; 김지후(강수리영) following beside him in the middle-right of the frame, midground, moves toward open gap between the goats; open gap between the goats in the lower-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 검은 염소 떼 (Dozens gathered inside the pen) — Goats flank both sides of the camera-facing gap, with partial bodies crossing the frame edges; used as Forms living foreground and lateral occlusion around the pair's narrow passage; 염소 떼 사이 빈 공간 (Temporarily clear after the village youths passed) — The opening runs from the lower center toward the approaching pair; used as Defines the precise route into which both characters are stepping.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low nighttime visibility inside the goat enclosure emphasizes dark silhouettes and restrained facial detail.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s animal-bone necklace remains on as she and Yeosang follow through the goat enclosure.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face); 김여상 (Korean man in his mid-40s, dark hair, ordinary mature face); 검은 염소 (coarse black hair, narrow muzzle, curved horns) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 신경을 곤두세운 채 검은 염소 떼 사이 빈 공간으로 조심스레 한쪽 발을 내디딘 김여상과 김지후(강수리영)의 모습.\n\nLOCATION (lock): The open-sided goat pen deep in the forest, among tightly gathered black goats. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Dolly in from waist height through the established open gap between the goats, remaining about two paces ahead of 김여상 and 김지후(강수리영). Frame them in a cautious full-body two-shot as each lowers their center of gravity and places one foot into the gap, while black goats form irregular foreground and side barriers without blocking their route.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김여상 testing the route in the middle-left of the frame, midground, moves toward open gap between the goats; 김지후(강수리영) following beside him in the middle-right of the frame, midground, moves toward open gap between the goats; open gap between the goats in the lower-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 검은 염소 떼 (Dozens gathered inside the pen) — Goats flank both sides of the camera-facing gap, with partial bodies crossing the frame edges; used as Forms living foreground and lateral occlusion around the pair's narrow passage; 염소 떼 사이 빈 공간 (Temporarily clear after the village youths passed) — The opening runs from the lower center toward the approaching pair; used as Defines the precise route into which both characters are stepping.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low nighttime visibility inside the goat enclosure emphasizes dark silhouettes and restrained facial detail.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s animal-bone necklace remains on as she and Yeosang follow through the goat enclosure.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face); 김여상 (Korean man in his mid-40s, dark hair, ordinary mature face); 검은 염소 (coarse black hair, narrow muzzle, curved horns) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 신경을 곤두세운 채 검은 염소 떼 사이 빈 공간으로 조심스레 한쪽 발을 내디딘 김여상과 김지후(강수리영)의 모습.\n\nLOCATION (lock): The open-sided goat pen deep in the forest, among tightly gathered black goats. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Dolly in from waist height through the established open gap between the goats, remaining about two paces ahead of 김여상 and 김지후(강수리영). Frame them in a cautious full-body two-shot as each lowers their center of gravity and places one foot into the gap, while black goats form irregular foreground and side barriers without blocking their route.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김여상 testing the route in the middle-left of the frame, midground, moves toward open gap between the goats; 김지후(강수리영) following beside him in the middle-right of the frame, midground, moves toward open gap between the goats; open gap between the goats in the lower-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 검은 염소 떼 (Dozens gathered inside the pen) — Goats flank both sides of the camera-facing gap, with partial bodies crossing the frame edges; used as Forms living foreground and lateral occlusion around the pair's narrow passage; 염소 떼 사이 빈 공간 (Temporarily clear after the village youths passed) — The opening runs from the lower center toward the approaching pair; used as Defines the precise route into which both characters are stepping.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low nighttime visibility inside the goat enclosure emphasizes dark silhouettes and restrained facial detail.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s animal-bone necklace remains on as she and Yeosang follow through the goat enclosure.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face); 김여상 (Korean man in his mid-40s, dark hair, ordinary mature face); 검은 염소 (coarse black hair, narrow muzzle, curved horns) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S31sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S31sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:813104>"
    },
    {
     "label": "CHARACTER REFERENCE — 김여상: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:990143>"
    },
    {
     "label": "CHARACTER REFERENCE — 검은 염소: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:892425>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L27B02.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:813104>"
    },
    {
     "label": "CHARACTER REFERENCE — 김여상: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:990143>"
    },
    {
     "label": "CHARACTER REFERENCE — 검은 염소: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:892425>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "구도와 낮춘 무게중심 연기는 지시에 근접하나 김여상 의상이 레퍼런스의 녹색 우비와 전혀 다르고 자세가 과장되게 쭈그려 부자연스러우며, 전경 염소가 과대하게 확대되어 실제 스케일 규정을 흔든다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "레퍼런스 장소·두 인물 의상·뼈 목걸이가 모두 정확하고 좌우 염소 벽 사이 하단 중앙 통로로 조심스레 발을 내딛는 야간 풀샷 2인 구도를 지시대로 실현했다(무게중심 낮춤이 약한 점만 감점)."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.75,
      "B": 1.667
     },
     "adjusted": {
      "A": 1.75,
      "B": 1.667
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "A"
     },
     "agreed": false
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "구도와 낮춘 무게중심 연기는 지시에 근접하나 김여상 의상이 레퍼런스의 녹색 우비와 전혀 다르고 자세가 과장되게 쭈그려 부자연스러우며, 전경 염소가 과대하게 확대되어 실제 스케일 규정을 흔든다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "레퍼런스 장소·두 인물 의상·뼈 목걸이가 모두 정확하고 좌우 염소 벽 사이 하단 중앙 통로로 조심스레 발을 내딛는 야간 풀샷 2인 구도를 지시대로 실현했다(무게중심 낮춤이 약한 점만 감점)."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "readings": [
     {
      "label": "A",
      "direction": "두 인물 모두 카메라 쪽(전방 하단)으로 걸어오며 시선은 발밑 통로와 전방 아래를 향한다. 김여상은 화면 중앙 좌측에서 고개를 숙여 발 디딜 곳을 보고, 김지후는 중앙 우측에서 전방을 응시한다. 두 사람의 진행 방향이 하단 중앙의 빈 통로와 정확히 일치하며, 카메라는 그 통로 안에서 두 사람보다 두어 걸음 앞에 위치한 시점이다.",
      "built_space": "레퍼런스의 개방형 축사가 그대로 재현됐다 — 목조 기둥열, 파형 강판 지붕, 채광 패널, 좌측 나무·철망 칸막이, 중앙 뒤편의 개방 통로와 숲. 지붕 구조와 기둥 간격, 바닥 흙길 자국이 레퍼런스와 부합. 카메라는 허리 높이보다 다소 높은 눈높이에 가까우나 축사 내부 통로 축을 정확히 따른다. 염소 떼가 좌우 양쪽에 늘어서 측면 차폐를 형성하고 중앙 통로는 비어 있다.",
      "entities": "김여상: 40대 중반 한국 남성, 짧은 검은 머리, 레퍼런스와 동일한 녹색 방수 재킷·바지·검은 장화 — 일치. 김지후: 20대 초반 한국 여성, 긴 검은 생머리, 갈색 워크재킷·올리브 이너·짙은 카고팬츠·워크부츠 — 일치(모자는 없음). 목에 밝은 색 뼈 펜던트 목걸이가 보여 CARRIED STATE 충족. 염소: 검은 거친 털, 좁은 주둥이, 굽은 뿔 — 레퍼런스와 부합, 수십 마리. 추가 인물 없음.",
      "hard_violations": [],
      "physics": "두 사람 모두 두 발이 지면에 접지되어 걷는 중; 김여상은 왼발이 앞으로 나가며 체중이 뒤발에 실린 자연스러운 보행, 김지후도 한 발 내딛는 중간 자세. 공중에 뜬 신체나 지지 없는 물체 없음. 염소들은 모두 네 발로 서 있고 바닥 그림자가 접지를 확인시킨다. 다만 '무게중심을 낮춘' 신중한 자세는 다소 약해 평범한 보행에 가깝다."
     },
     {
      "label": "B",
      "direction": "김여상은 화면 좌중앙에서 몸을 낮추고 두 팔을 벌린 채 카메라 쪽 하단으로 향하지만 시선은 아래 왼쪽 바닥을 본다. 김지후는 우중앙에서 역시 전방 하단을 보며 이동. 진행 방향은 하단 중앙의 빈 공간과 대체로 일치. 다만 두 사람 사이 간격이 넓어 '나란히 따르는' 관계가 느슨하다.",
      "built_space": "개방형 축사 구조는 유지되나 지붕 채광 패널·기둥 배치가 레퍼런스와 다소 다르고 뒤쪽 개구부가 어둡게 막힌 듯 보인다. 좌우 및 전경에 염소가 배치되어 측면 차폐를 만들지만, 전경 우측 하단의 염소가 인물보다 훨씬 거대하게 확대되어 실제 물리적 스케일(사람 대 염소)이 붕괴됐다 — 어깨높이가 프레임 절반을 차지한다. 카메라 높이는 허리보다 높은 편.",
      "entities": "김여상: 40대 남성이나 복장이 레퍼런스의 녹색 방수복이 아니라 회색빛 재킷·짙은 바지로 어긋남. 김지후: 20대 여성, 긴 검은 머리, 갈색 재킷·목에 흰 뼈 펜던트 확인 — 대체로 일치. 염소: 검은 털·굽은 뿔로 종은 맞으나 전경 개체 크기가 비현실적. 추가 인물 없음.",
      "hard_violations": [
       "전경 염소가 인물 대비 비현실적으로 거대하게 확대되어 실제 물리적 스케일이 붕괴됨(배경 요소를 전경 존재로 확대하지 말라는 지시 위반)"
      ],
      "physics": "두 인물 모두 발이 지면에 있고 무릎을 굽혀 무게중심을 낮춘 자세 — 김여상은 스쿼트에 가까운 과장된 자세로 접지는 확인되나 걷는 동작으로는 다소 부자연스럽다. 공중 부양 신체는 없음. 염소들은 접지되어 있으나 전경 개체는 원근상 크기가 맞지 않아 무게감이 어색하다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "레퍼런스 축사·의상·뼈목걸이가 정확히 일치하고, 염소가 좌우 측면 차폐를 만든 중앙 빈 통로로 두 인물이 나란히 발을 내딛는 야간 와이드 투샷 구도를 지시대로 실현했다; 다만 '무게중심을 낮춘' 신중함이 다소 약하다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "조심스레 무게중심을 낮춘 동작은 잘 살렸으나 전경 염소가 인물보다 거대하게 확대돼 실제 스케일 규정을 정면으로 어겼고, 김여상의 녹색 방수복 의상도 레퍼런스와 불일치한다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "레퍼런스 축사·의상·뼈목걸이가 정확히 일치하고, 염소가 좌우 측면 차폐를 만든 중앙 빈 통로로 두 인물이 나란히 발을 내딛는 야간 와이드 투샷 구도를 지시대로 실현했다; 다만 '무게중심을 낮춘' 신중함이 다소 약하다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "조심스레 무게중심을 낮춘 동작은 잘 살렸으나 전경 염소가 인물보다 거대하게 확대돼 실제 스케일 규정을 정면으로 어겼고, 김여상의 녹색 방수복 의상도 레퍼런스와 불일치한다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "두 인물 모두 카메라 쪽(전방 하단)으로 걸어오며 시선은 발밑 통로와 전방 아래를 향한다. 김여상은 화면 중앙 좌측에서 고개를 숙여 발 디딜 곳을 보고, 김지후는 중앙 우측에서 전방을 응시한다. 두 사람의 진행 방향이 하단 중앙의 빈 통로와 정확히 일치하며, 카메라는 그 통로 안에서 두 사람보다 두어 걸음 앞에 위치한 시점이다.",
      "built_space": "레퍼런스의 개방형 축사가 그대로 재현됐다 — 목조 기둥열, 파형 강판 지붕, 채광 패널, 좌측 나무·철망 칸막이, 중앙 뒤편의 개방 통로와 숲. 지붕 구조와 기둥 간격, 바닥 흙길 자국이 레퍼런스와 부합. 카메라는 허리 높이보다 다소 높은 눈높이에 가까우나 축사 내부 통로 축을 정확히 따른다. 염소 떼가 좌우 양쪽에 늘어서 측면 차폐를 형성하고 중앙 통로는 비어 있다.",
      "entities": "김여상: 40대 중반 한국 남성, 짧은 검은 머리, 레퍼런스와 동일한 녹색 방수 재킷·바지·검은 장화 — 일치. 김지후: 20대 초반 한국 여성, 긴 검은 생머리, 갈색 워크재킷·올리브 이너·짙은 카고팬츠·워크부츠 — 일치(모자는 없음). 목에 밝은 색 뼈 펜던트 목걸이가 보여 CARRIED STATE 충족. 염소: 검은 거친 털, 좁은 주둥이, 굽은 뿔 — 레퍼런스와 부합, 수십 마리. 추가 인물 없음.",
      "hard_violations": [],
      "physics": "두 사람 모두 두 발이 지면에 접지되어 걷는 중; 김여상은 왼발이 앞으로 나가며 체중이 뒤발에 실린 자연스러운 보행, 김지후도 한 발 내딛는 중간 자세. 공중에 뜬 신체나 지지 없는 물체 없음. 염소들은 모두 네 발로 서 있고 바닥 그림자가 접지를 확인시킨다. 다만 '무게중심을 낮춘' 신중한 자세는 다소 약해 평범한 보행에 가깝다."
     },
     {
      "label": "A",
      "direction": "김여상은 화면 좌중앙에서 몸을 낮추고 두 팔을 벌린 채 카메라 쪽 하단으로 향하지만 시선은 아래 왼쪽 바닥을 본다. 김지후는 우중앙에서 역시 전방 하단을 보며 이동. 진행 방향은 하단 중앙의 빈 공간과 대체로 일치. 다만 두 사람 사이 간격이 넓어 '나란히 따르는' 관계가 느슨하다.",
      "built_space": "개방형 축사 구조는 유지되나 지붕 채광 패널·기둥 배치가 레퍼런스와 다소 다르고 뒤쪽 개구부가 어둡게 막힌 듯 보인다. 좌우 및 전경에 염소가 배치되어 측면 차폐를 만들지만, 전경 우측 하단의 염소가 인물보다 훨씬 거대하게 확대되어 실제 물리적 스케일(사람 대 염소)이 붕괴됐다 — 어깨높이가 프레임 절반을 차지한다. 카메라 높이는 허리보다 높은 편.",
      "entities": "김여상: 40대 남성이나 복장이 레퍼런스의 녹색 방수복이 아니라 회색빛 재킷·짙은 바지로 어긋남. 김지후: 20대 여성, 긴 검은 머리, 갈색 재킷·목에 흰 뼈 펜던트 확인 — 대체로 일치. 염소: 검은 털·굽은 뿔로 종은 맞으나 전경 개체 크기가 비현실적. 추가 인물 없음.",
      "hard_violations": [
       "전경 염소가 인물 대비 비현실적으로 거대하게 확대되어 실제 물리적 스케일이 붕괴됨(배경 요소를 전경 존재로 확대하지 말라는 지시 위반)"
      ],
      "physics": "두 인물 모두 발이 지면에 있고 무릎을 굽혀 무게중심을 낮춘 자세 — 김여상은 스쿼트에 가까운 과장된 자세로 접지는 확인되나 걷는 동작으로는 다소 부자연스럽다. 공중 부양 신체는 없음. 염소들은 접지되어 있으나 전경 개체는 원근상 크기가 맞지 않아 무게감이 어색하다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 10,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 10,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "구도와 낮춘 무게중심 연기는 지시에 근접하나 김여상 의상이 레퍼런스의 녹색 우비와 전혀 다르고 자세가 과장되게 쭈그려 부자연스러우며, 전경 염소가 과대하게 확대되어 실제 스케일 규정을 흔든다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "레퍼런스 장소·두 인물 의상·뼈 목걸이가 모두 정확하고 좌우 염소 벽 사이 하단 중앙 통로로 조심스레 발을 내딛는 야간 풀샷 2인 구도를 지시대로 실현했다(무게중심 낮춤이 약한 점만 감점)."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L27B02.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:813104>"
   },
   {
    "label": "CHARACTER REFERENCE — 김여상: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:990143>"
   },
   {
    "label": "CHARACTER REFERENCE — 검은 염소: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:892425>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "두 인물이 중심을 낮추고 조심스럽게 한 발을 내딛어야 하나, 현재는 허리를 곧게 펴고 평범하게 서서 걷는 자세입니다.",
     "fix_en": "Redraw Kim Yeosang and Kim Jihu with a lowered center of gravity, bending their knees and cautiously extending one foot forward into the gap as if testing the ground. Preserve their exact identities, screen positions, their complete outfits including Jihu's bone necklace, the surrounding black goats, the goat pen environment, the night lighting, and the wide-shot framing exactly.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 5,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw Kim Yeosang and Kim Jihu with a lowered center of gravity, bending their knees and cautiously extending one foot forward into the gap as if testing the ground. Preserve their exact identities, screen positions, their complete outfits including Jihu's bone necklace, the surrounding black goats, the goat pen environment, the night lighting, and the wide-shot framing exactly.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "로케이션·인물·야간 톤은 충실하지만 '무게중심을 낮추고 한쪽 발을 조심스레 내딛는' 샷 텍스트의 핵심 동작이 전혀 없이 평범한 보행이라 스테이징 충실도에서 밀린다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 세트·의상·목걸이·야간 조도를 유지한 채 두 인물이 무릎을 굽혀 중심을 낮추고 한 발을 빈 통로로 내미는 지정 동작을 정확히 구현했고, 다른 요소를 바꾸지 않아 수리 비용이 거의 없다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.528,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.528,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "로케이션·인물·야간 톤은 충실하지만 '무게중심을 낮추고 한쪽 발을 조심스레 내딛는' 샷 텍스트의 핵심 동작이 전혀 없이 평범한 보행이라 스테이징 충실도에서 밀린다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 세트·의상·목걸이·야간 조도를 유지한 채 두 인물이 무릎을 굽혀 중심을 낮추고 한 발을 빈 통로로 내미는 지정 동작을 정확히 구현했고, 다른 요소를 바꾸지 않아 수리 비용이 거의 없다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 인물 모두 시선을 프레임 하단 중앙, 즉 염소 떼 사이 빈 통로(카메라 쪽으로 열린 길)로 내리깔고 있다. 김여상은 왼쪽 중간, 김지후는 오른쪽 중간에 서서 몸통과 발끝이 모두 카메라(=통로 진행 방향)를 향한다. 시선이 실제로 발을 딛을 지면에 착지하므로 '신경을 곤두세운 채' 발밑을 살피는 지시와 일치.",
      "built_space": "레퍼런스 축사 구조 재현: 목재 기둥열, 골함석 지붕과 상부 채광 슬랫, 좌측 철망·목재 칸막이 게이트, 정면 개구부 너머 숲과 흙길이 모두 위치·재질 그대로. 지붕 트러스 개수와 기둥 간격도 레퍼런스와 대체로 부합. 전경 좌우에 염소 몸통이 프레임 가장자리를 잘라 측면 차폐를 형성하고 중앙 하단에 흙 통로가 비어 있다.",
      "entities": "김여상 — 40대 한국 남성, 짧은 검은 머리, 녹색 방수 재킷+가슴받이 바지+고무장화로 레퍼런스 의상 일치. 김지후 — 20대 초반 한국 여성, 긴 검은 생머리(모자 없음), 갈색 워크재킷·올리브 이너·짙은 카고팬츠·워크부츠, 가슴에 밝은 뼈(뿔) 모양 펜던트 목걸이가 뚜렷 — CARRIED STATE 충족. 염소들은 거친 검은 털, 좁은 주둥이, 굽은 뿔로 레퍼런스와 일치, 수십 마리 규모.",
      "hard_violations": [],
      "physics": "두 사람 모두 무릎을 굽혀 중심을 낮춘 반쯤 웅크린 자세. 김여상은 뒷발이 땅에 붙고 앞발을 통로 흙바닥에 내려놓는 중, 손은 균형용으로 무릎 앞에 벌어져 있다. 김지후도 한쪽 발을 앞으로 내밀고 반대 발로 체중을 지지 — 무게중심 이동이 해부학적으로 성립. 공중에 뜬 신체나 무지지 물체 없음."
     },
     {
      "label": "B",
      "direction": "두 사람 모두 시선을 아래·전방으로 내리고 있으나 몸은 직립 상태로 그냥 걸어 나온다. 김여상의 시선은 발밑을 향하고 김지후는 정면 아래를 본다. 진행 방향은 카메라 쪽 통로로 맞지만 '조심스레 한쪽 발을 내디디는' 조준점(발이 놓일 빈 공간)에 대한 신체적 집중은 약하다.",
      "built_space": "동일한 축사 구조를 재현하고 오히려 지붕·기둥·상부 채광 슬랫이 더 넓게 보인다. 좌측 철망 게이트, 정면 개구부와 숲, 흙 통로 모두 레퍼런스와 일치. 구조상 오류 없음.",
      "entities": "김여상 — 녹색 방수복·장화, 안에 남색 셔츠가 레퍼런스대로 드러나 있어 의상 일치도 소폭 더 높음. 김지후 — 긴 검은 머리, 갈색 재킷, 목에 밝은 뼈 펜던트 목걸이 유지. 염소들은 좌우 전경에 배치되어 있으나 A보다 인물에서 멀고 통로가 넓어 '빽빽이 모인' 밀집감은 덜하다.",
      "hard_violations": [],
      "physics": "두 사람 모두 완전한 직립 보행 자세. 한쪽 발이 지면에 닿고 다른 발이 앞으로 나가는 평범한 걷기이며 지지는 명확하다. 다만 계약이 요구한 '무게중심을 낮추고 한쪽 발을 빈 공간에 조심스레 내딛는' 순간은 재현되지 않음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "무릎을 굽혀 중심을 낮추고 한쪽 발을 통로에 내딛는 '조심스러운' 순간과 좌우 염소의 전경 차폐가 계약 그대로 구현되어 스테이징 충실도가 가장 높다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "김여상의 셔츠 등 의상 디테일은 미세하게 나아졌지만, 계약의 핵심인 '중심을 낮춘 채 한쪽 발을 내딛는' 자세를 평범한 직립 보행으로 바꾸고 염소 밀집·전경 차폐감까지 약화시켜 얻은 것보다 잃은 것이 크다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "무릎을 굽혀 중심을 낮추고 한쪽 발을 통로에 내딛는 '조심스러운' 순간과 좌우 염소의 전경 차폐가 계약 그대로 구현되어 스테이징 충실도가 가장 높다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "김여상의 셔츠 등 의상 디테일은 미세하게 나아졌지만, 계약의 핵심인 '중심을 낮춘 채 한쪽 발을 내딛는' 자세를 평범한 직립 보행으로 바꾸고 염소 밀집·전경 차폐감까지 약화시켜 얻은 것보다 잃은 것이 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "두 인물 모두 시선을 프레임 하단 중앙, 즉 염소 떼 사이 빈 통로(카메라 쪽으로 열린 길)로 내리깔고 있다. 김여상은 왼쪽 중간, 김지후는 오른쪽 중간에 서서 몸통과 발끝이 모두 카메라(=통로 진행 방향)를 향한다. 시선이 실제로 발을 딛을 지면에 착지하므로 '신경을 곤두세운 채' 발밑을 살피는 지시와 일치.",
      "built_space": "레퍼런스 축사 구조 재현: 목재 기둥열, 골함석 지붕과 상부 채광 슬랫, 좌측 철망·목재 칸막이 게이트, 정면 개구부 너머 숲과 흙길이 모두 위치·재질 그대로. 지붕 트러스 개수와 기둥 간격도 레퍼런스와 대체로 부합. 전경 좌우에 염소 몸통이 프레임 가장자리를 잘라 측면 차폐를 형성하고 중앙 하단에 흙 통로가 비어 있다.",
      "entities": "김여상 — 40대 한국 남성, 짧은 검은 머리, 녹색 방수 재킷+가슴받이 바지+고무장화로 레퍼런스 의상 일치. 김지후 — 20대 초반 한국 여성, 긴 검은 생머리(모자 없음), 갈색 워크재킷·올리브 이너·짙은 카고팬츠·워크부츠, 가슴에 밝은 뼈(뿔) 모양 펜던트 목걸이가 뚜렷 — CARRIED STATE 충족. 염소들은 거친 검은 털, 좁은 주둥이, 굽은 뿔로 레퍼런스와 일치, 수십 마리 규모.",
      "hard_violations": [],
      "physics": "두 사람 모두 무릎을 굽혀 중심을 낮춘 반쯤 웅크린 자세. 김여상은 뒷발이 땅에 붙고 앞발을 통로 흙바닥에 내려놓는 중, 손은 균형용으로 무릎 앞에 벌어져 있다. 김지후도 한쪽 발을 앞으로 내밀고 반대 발로 체중을 지지 — 무게중심 이동이 해부학적으로 성립. 공중에 뜬 신체나 무지지 물체 없음."
     },
     {
      "label": "A",
      "direction": "두 사람 모두 시선을 아래·전방으로 내리고 있으나 몸은 직립 상태로 그냥 걸어 나온다. 김여상의 시선은 발밑을 향하고 김지후는 정면 아래를 본다. 진행 방향은 카메라 쪽 통로로 맞지만 '조심스레 한쪽 발을 내디디는' 조준점(발이 놓일 빈 공간)에 대한 신체적 집중은 약하다.",
      "built_space": "동일한 축사 구조를 재현하고 오히려 지붕·기둥·상부 채광 슬랫이 더 넓게 보인다. 좌측 철망 게이트, 정면 개구부와 숲, 흙 통로 모두 레퍼런스와 일치. 구조상 오류 없음.",
      "entities": "김여상 — 녹색 방수복·장화, 안에 남색 셔츠가 레퍼런스대로 드러나 있어 의상 일치도 소폭 더 높음. 김지후 — 긴 검은 머리, 갈색 재킷, 목에 밝은 뼈 펜던트 목걸이 유지. 염소들은 좌우 전경에 배치되어 있으나 A보다 인물에서 멀고 통로가 넓어 '빽빽이 모인' 밀집감은 덜하다.",
      "hard_violations": [],
      "physics": "두 사람 모두 완전한 직립 보행 자세. 한쪽 발이 지면에 닿고 다른 발이 앞으로 나가는 평범한 걷기이며 지지는 명확하다. 다만 계약이 요구한 '무게중심을 낮추고 한쪽 발을 빈 공간에 조심스레 내딛는' 순간은 재현되지 않음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S31sh3__bgfirst_bg.png",
   "bg_asset_id": "3bb5c2ec-11db-4118-8872-a7d74b2f31aa",
   "bg_record_key": "S31sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S32sh12": {
  "input_fingerprint": "f1a8ae59abe81817",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 유리창 밖에서 터진 핸드폰 카메라 플래시 불빛이 도축장 안을 번쩍 밝힌 순간.\n\nLOCATION (lock): The slaughterhouse work floor facing a broken, dusty window through which a phone flash bursts. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Hold static at chest height from behind and to one side of the gathering point, facing the broken window across the room at the very first instant of the phone-camera flash. The window sits near the upper center as the flash abruptly exposes the hanging tools, bucket, and rough interior before any human turn is completed or clearly framed.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: broken window containing the phone-camera flash in the upper-center of the frame, background; slaughter tools revealed by the flash in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 깨진 유리창 (Glass broken; dust and cobwebs present) — Broken glass, dust, and cobwebs remain visible around the opening, with the exterior phone flash seen through it; used as Primary background anchor through which the phone-camera flash enters; 벽에 매달린 칼과 걸개 갈고리 (Hanging in rows) — Their hanging faces, edges, and hooked profiles face into the room; used as The flash reveals their rows across the interior wall; 피받이 양동이 (Placed beneath the slaughter area) — Seen from the side with part of its opening visible; used as Provides a low spatial reference beneath the flash-revealed tools; 나무로 얼기설기 만든 가건물 내부 (Used as a slaughterhouse) — The interior wall and its tool-covered side face the camera; used as Defines the enclosure suddenly exposed by the flash.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: A sudden phone-camera flash from outside the broken window momentarily overwhelms the previously dark interior with hard, high-contrast illumination.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu still wears her necklace as Yeosang’s phone flash exposes the blood-covered slaughter scene.\n\nPEOPLE: the SHOT TEXT alone decides who is visible in this shot. People known to appear somewhere in this scene: 마을 청년 #2 (Korean young man, short dark hair, lean youthful face); 마을 청년 #1 (Korean young man, short dark hair, broad youthful face). That list is scene-level, not a cast list for this frame — it may name someone this shot does not show, and it may omit someone this shot does show. If the shot text names a person who is not on the list, draw that person exactly as the shot text describes them; the list does not override the shot text. Never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "카메라는 실내에서 창을 정면으로 바라보고, 상단 중앙 깨진 구멍으로 외부 핸드폰 플래시가 들어온다. 방향 자체는 맞으나 화면 앞쪽에 다섯 명 이상의 남자 뒤통수가 좌우로 늘어서 창을 응시한다. 시선 타깃은 창 = 일치.",
    "built_space": "가건물 내부이지만 레퍼런스의 콘크리트/석회벽·도살대·천장 등·냉장고 문·매달린 고기 등 고정 요소가 대부분 사라지고 판자벽 소품 세트로 재구성됨. 칼 걸이 한 줄, 갈고리, 양동이는 존재. 도살대(작업대)와 천장 조명, 시신은 사라짐.",
    "entities": "샷 텍스트가 부르지 않은 인물 5명 이상이 전경을 채운다(장면 목록엔 청년 2명뿐). 창밖 인물의 손·플래시는 보이나, 필수 배경 요소인 도살대와 이전 컷의 인물/시신은 부재.",
    "physics": "인물들은 바닥에 서 있어 지지는 정상. 플래시 광원은 얼굴 정면을 비추지 못하고 인물 실루엣만 남겨 광 확산이 다소 비물리적이나 치명적이진 않음. 벽 중앙에 정체불명의 청록색 렌즈 플레어 점이 표면에 붙어 있음.",
    "hard_violations": [
     "샷 텍스트가 지정하지 않은 다수 인물(5명 이상) 추가 — 발명된 인물",
     "레퍼런스가 고정한 장소(콘크리트 벽·도살대·천장등·냉장문·시신)를 판자벽 신규 공간으로 대체"
    ]
   },
   {
    "label": "B",
    "direction": "카메라가 실내 측면에서 창을 바라보고 상단 중앙 깨진 구멍에서 플래시가 터진다. 우측 전경 남자 뒤통수/옆얼굴이 창 쪽을 보고 있어 시선 타깃은 창 = 일치.",
    "built_space": "칼 열, 갈고리 열, 바닥의 양동이가 모두 벽에 존재하고 창은 상단 중앙. 다만 레퍼런스의 핵심 고정 요소인 나무 도살대, 천장 백열등, 매달린 고기, 냉장고 문, 누워 있는 시신이 전부 사라졌고 좁은 판자 코너로 바뀜. 양동이는 피가 아닌 빈 금속통.",
    "entities": "전경 인물 1명(짧은 머리 한국 남성)은 이전 컷 인물의 재킷과 유사하나 정면 확인 불가. 창밖 촬영자 실루엣 존재. 시신·도살대 등 락된 요소 누락.",
    "physics": "인물은 바닥에 서 있고 지지 정상. 플래시 하이라이트가 인물 뒤통수와 벽에 맞게 떨어짐. 벽 중앙에 렌즈 플레어 색점이 표면에 붙은 듯 보이는 결함 있음.",
    "hard_violations": []
   },
   {
    "label": "C",
    "direction": "카메라는 도살 지점 뒤·측면 가슴 높이에서 방 건너 창을 향하고, 창은 상단 중앙에서 플래시가 터진다. 창밖 인물이 폰을 든 채 보이고, 도살대 위 인물의 시선은 창/카메라 쪽으로 놀란 듯 향한다. 전경 남자는 칼을 내린 채 아직 몸을 돌리는 중 — '아직 인간의 turn이 완료되지 않은' 순간과 일치.",
    "built_space": "레퍼런스와 동일한 콘크리트 벽, 나무 도살대, 천장 백열등, 좌측 냉장문과 매달린 고기, 우측 링 갈고리·쇠사슬·철제 문, 피받이 양동이가 모두 정확히 계승됨. 칼과 갈고리 열이 벽에 정면으로 배치. 창이 새로 상단 중앙에 추가되어 깨진 유리·거미줄·먼지가 보임.",
    "entities": "인물은 이전 컷의 두 사람(칼 든 남자, 도살대 위 인물)뿐 — 새 인물 추가 없음. 창밖에는 폰을 든 손과 얼굴 실루엣만. 피받이 양동이는 피로 가득 찬 동일 소품.",
    "physics": "칼 든 남자는 두 발이 바닥에 있고 손이 칼자루를 쥔 상태로 지지 명확. 도살대 위 인물은 상체를 대에 얹은 채 팔로 지지하며 고개를 든 자세로, 물리적으로 성립. 부유하는 물체 없음.",
    "hard_violations": []
   }
  ],
  "totals": {
   "A": 3,
   "B": 6,
   "C": 9
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "샷 텍스트에 없는 다섯 명 이상의 인물을 전경에 발명해 넣었고, 레퍼런스가 락한 도살장 구조(도살대·천장등·시신)를 판자벽 세트로 바꿔 장소 락을 파기했다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "인물 추가 없이 창·플래시·칼열·갈고리·양동이를 정확히 배치했으나, 레퍼런스가 락한 도살대·천장등·시신·매달린 고기가 전부 사라져 장소 연속성이 약하고 프레임도 좁은 코너로 축소됐다."
   },
   {
    "label": "C",
    "score": 9,
    "verdict_ko": "레퍼런스 공간(도살대·천장등·냉장문·양동이·갈고리)을 그대로 유지한 채 가슴 높이 측후방 와이드로 상단 중앙 깨진 창의 플래시 순간을 포착했고, 인물 추가 없이 아직 돌아서지 못한 반응 순간을 물리적으로 타당하게 담았다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S32sh10_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "[Critical] 서 있는 남자의 복부 앞쪽 탁자 위에 어두운 소매를 입은 정체불명의 세 번째 손이 작은 칼을 쥐고 있습니다.",
     "fix_en": "Remove the extra hand holding a small knife from the table in front of the standing man's abdomen, replacing it with the empty, blood-stained surface of the table. Preserve the standing man, the man on the table, the cleaver, the lighting, the table, and the background framing exactly as they are."
    },
    {
     "issue_ko": "[Critical] 연속성 오류: 레퍼런스에서 서 있는 남자는 두 손으로 식칼을 머리 위로 높이 들고 있었으나, 이 사진에서는 한 손으로 식칼을 아래로 내리고 있어 자세가 일치하지 않습니다.",
     "fix_en": "Modify the standing man's arms so he is holding the large cleaver high above his head with both hands, matching his previous stance. Preserve the camera angle, his clothing, the other man's pose, the lighting, and the background framing exactly.",
     "needs_regeneration": true
    },
    {
     "issue_ko": "[Critical] 연속성 오류: 레퍼런스에서 탁자 위의 남자는 등을 대고 누워 있었으나, 이 사진에서는 상체를 일으키고 앉아 있어 자세가 일치하지 않습니다.",
     "fix_en": "Redraw the man on the table so he is lying flat on his back with his head resting on the table. Preserve the standing man's pose, the lighting, the table, and the background framing exactly.",
     "needs_regeneration": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 2,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the extra hand holding a small knife from the table in front of the standing man's abdomen, replacing it with the empty, blood-stained surface of the table. Preserve the standing man, the man on the table, the cleaver, the lighting, the table, and the background framing exactly as they are.\n- Modify the standing man's arms so he is holding the large cleaver high above his head with both hands, matching his previous stance. Preserve the camera angle, his clothing, the other man's pose, the lighting, and the background framing exactly.\n- Redraw the man on the table so he is lying flat on his back with his head resting on the table. Preserve the standing man's pose, the lighting, the table, and the background framing exactly.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 도살대 뒤·측면에서 방 건너편 깨진 창을 향하고, 창은 프레임 상단 중앙에 위치해 규정된 축과 일치. 창 밖 인물의 핸드폰 플래시가 실내를 향해 정면으로 터지고 있음. 전경 남자(청재킷)는 등을 보이며 도살대 쪽을 향하고 시선은 아직 창으로 돌아가지 않은 상태로 '아직 고개를 돌리기 전' 순간에 부합. 그러나 문제는 도살대 위에 있어야 할 인물이 서 있는 상태로 대(臺) 위에 상체를 숙이고 카메라/전경 남자 쪽을 응시함 — 이전 스틸에서 누워 있던 인물이 일어나 다른 자세·다른 위치를 취함.",
      "built_space": "판자·콘크리트 벽, 좌측 냉장고형 철문, 매달린 갈고리와 사슬, 벽에 늘어선 칼 열, 우측 하단 피받이 양동이 — 이전 스틸의 고정 요소가 대체로 유지됨. 창은 상단 중앙, 깨진 유리와 거미줄 텍스처 존재. 다만 이전 스틸에서는 천장 조명이 켜져 있었고 여기서도 켜져 있어 '이전까지 어두웠던 실내'라는 조명 설정과는 다소 어긋남(양 후보 공통). 도살대 배치와 양동이 위치는 레퍼런스와 일치.",
      "entities": "한국인 젊은 남성 2인. 전경 남성은 이전 스틸의 청재킷 남자(칼을 든 인물)와 의상 일치, 손에 큰 정육도를 낮게 들고 있음(이전 스틸에서는 머리 위로 치켜듦 — 동작이 진행된 것으로 볼 여지는 있으나 '움직일 수 없는 인물은 그대로'라는 조항은 누워 있던 피해자에 해당). 두 번째 인물은 이전 스틸에서 도살대 위에 누워 있던 피해자여야 하는데, 여기서는 상체를 들고 대 위에 몸을 걸친 채 눈을 뜨고 응시 — 정체·상태가 바뀜. 창밖 인물은 실루엣으로 핸드폰을 든 손과 얼굴 일부만 보임(추가 인물이 아니라 샷 텍스트가 요구한 플래시 주체).",
      "hard_violations": [
       "이전 스틸에서 도살대 위에 움직이지 못하는 상태로 누워 있던 인물이 상체를 일으켜 눈을 뜨고 응시하는 완전히 다른 자세·위치로 렌더링됨 — '움직일 수 없는 인물의 포즈·위치·접촉점 고정' 조항 위반"
      ],
      "physics": "전경 남자는 두 발로 바닥을 딛고 칼자루를 오른손으로 쥐고 있음 — 지지 정상. 두 번째 인물은 두 팔과 상체를 도살대 상판에 얹고 다리는 프레임 밖(전경 남자에 가림)으로, 지지 자체는 성립하나 이전 스틸의 사망/무력 상태와 모순되는 능동적 자세. 양동이는 바닥에 놓여 있고 피가 고여 광택 있음 — 물리적으로 타당."
     },
     {
      "label": "B",
      "direction": "카메라는 가슴 높이, 도살대 뒤·좌측에서 방 건너편 깨진 창을 향하고 창은 상단 중앙 — 규정 축과 프레임 레이아웃에 부합. 창 밖 실루엣의 핸드폰 플래시가 실내를 정면으로 비춤. 전경 남자는 창을 향해 몸을 반쯤 돌리고 양손을 들어올린 상태이며 시선은 창/플래시 쪽으로 향함. 벽의 칼 열과 갈고리, 피받이 흔적이 중앙 중경에 노출됨.",
      "built_space": "좌측 철문·복도, 상단 대들보와 사슬, 우측 갈고리 세트, 벽의 칼 열, 도살대 — 레퍼런스 고정 요소 유지. 다만 우측 하단 피받이 양동이가 사라지고 도살대만 보이며, 칼이 놓인 작은 보조 테이블이 새로 등장. 천장등 켜짐. 창의 깨진 형상·거미줄은 레퍼런스 톤과 일치.",
      "entities": "한국인 젊은 남성 2인. 전경 남성은 어두운 데님 재킷으로 이전 스틸 인물과 의상 일치, 짧은 검은 머리. 도살대 위 인물은 이전 스틸처럼 반듯이 누워 얼굴을 위로 향한 무력 상태로 유지되며 흙빛 상의도 일치 — 고정 프롭 조항 준수. 창밖 실루엣은 핸드폰을 든 손만 보임.",
      "hard_violations": [],
      "physics": "전경 남자는 두 발로 바닥을 딛고 서서 양팔을 위로 든 자세 — 지지 정상이며 놀라 손을 드는 순간으로 자연스러움(단 손에 들고 있던 정육도가 사라짐). 누운 인물은 도살대 상판 전체가 받치고 있고 머리·어깨·다리 접촉점이 명확. 공중에 뜬 물체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "창·플래시 배치와 칼 열·양동이 등 배경 계약은 잘 지켰고 칼을 든 상태도 유지했지만, 이전 스틸에서 도살대 위에 무력하게 누워 있던 인물을 상체를 일으켜 눈뜨고 응시하는 전혀 다른 자세로 바꾼 것이 '움직일 수 없는 인물 고정' 조항의 결정적 위반."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "누운 피해자의 포즈·위치·의상을 레퍼런스 그대로 고정했고 창 상단 중앙·플래시 순간·와이드 축을 충족 — 피받이 양동이 누락과 손에 든 정육도 소실은 계약이 요구하지 않은 손실이나, A의 인물 상태 변경보다 훨씬 가벼워 우세."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "창·플래시 배치와 칼 열·양동이 등 배경 계약은 잘 지켰고 칼을 든 상태도 유지했지만, 이전 스틸에서 도살대 위에 무력하게 누워 있던 인물을 상체를 일으켜 눈뜨고 응시하는 전혀 다른 자세로 바꾼 것이 '움직일 수 없는 인물 고정' 조항의 결정적 위반."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "누운 피해자의 포즈·위치·의상을 레퍼런스 그대로 고정했고 창 상단 중앙·플래시 순간·와이드 축을 충족 — 피받이 양동이 누락과 손에 든 정육도 소실은 계약이 요구하지 않은 손실이나, A의 인물 상태 변경보다 훨씬 가벼워 우세."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 도살대 뒤·측면에서 방 건너편 깨진 창을 향하고, 창은 프레임 상단 중앙에 위치해 규정된 축과 일치. 창 밖 인물의 핸드폰 플래시가 실내를 향해 정면으로 터지고 있음. 전경 남자(청재킷)는 등을 보이며 도살대 쪽을 향하고 시선은 아직 창으로 돌아가지 않은 상태로 '아직 고개를 돌리기 전' 순간에 부합. 그러나 문제는 도살대 위에 있어야 할 인물이 서 있는 상태로 대(臺) 위에 상체를 숙이고 카메라/전경 남자 쪽을 응시함 — 이전 스틸에서 누워 있던 인물이 일어나 다른 자세·다른 위치를 취함.",
      "built_space": "판자·콘크리트 벽, 좌측 냉장고형 철문, 매달린 갈고리와 사슬, 벽에 늘어선 칼 열, 우측 하단 피받이 양동이 — 이전 스틸의 고정 요소가 대체로 유지됨. 창은 상단 중앙, 깨진 유리와 거미줄 텍스처 존재. 다만 이전 스틸에서는 천장 조명이 켜져 있었고 여기서도 켜져 있어 '이전까지 어두웠던 실내'라는 조명 설정과는 다소 어긋남(양 후보 공통). 도살대 배치와 양동이 위치는 레퍼런스와 일치.",
      "entities": "한국인 젊은 남성 2인. 전경 남성은 이전 스틸의 청재킷 남자(칼을 든 인물)와 의상 일치, 손에 큰 정육도를 낮게 들고 있음(이전 스틸에서는 머리 위로 치켜듦 — 동작이 진행된 것으로 볼 여지는 있으나 '움직일 수 없는 인물은 그대로'라는 조항은 누워 있던 피해자에 해당). 두 번째 인물은 이전 스틸에서 도살대 위에 누워 있던 피해자여야 하는데, 여기서는 상체를 들고 대 위에 몸을 걸친 채 눈을 뜨고 응시 — 정체·상태가 바뀜. 창밖 인물은 실루엣으로 핸드폰을 든 손과 얼굴 일부만 보임(추가 인물이 아니라 샷 텍스트가 요구한 플래시 주체).",
      "hard_violations": [
       "이전 스틸에서 도살대 위에 움직이지 못하는 상태로 누워 있던 인물이 상체를 일으켜 눈을 뜨고 응시하는 완전히 다른 자세·위치로 렌더링됨 — '움직일 수 없는 인물의 포즈·위치·접촉점 고정' 조항 위반"
      ],
      "physics": "전경 남자는 두 발로 바닥을 딛고 칼자루를 오른손으로 쥐고 있음 — 지지 정상. 두 번째 인물은 두 팔과 상체를 도살대 상판에 얹고 다리는 프레임 밖(전경 남자에 가림)으로, 지지 자체는 성립하나 이전 스틸의 사망/무력 상태와 모순되는 능동적 자세. 양동이는 바닥에 놓여 있고 피가 고여 광택 있음 — 물리적으로 타당."
     },
     {
      "label": "B",
      "direction": "카메라는 가슴 높이, 도살대 뒤·좌측에서 방 건너편 깨진 창을 향하고 창은 상단 중앙 — 규정 축과 프레임 레이아웃에 부합. 창 밖 실루엣의 핸드폰 플래시가 실내를 정면으로 비춤. 전경 남자는 창을 향해 몸을 반쯤 돌리고 양손을 들어올린 상태이며 시선은 창/플래시 쪽으로 향함. 벽의 칼 열과 갈고리, 피받이 흔적이 중앙 중경에 노출됨.",
      "built_space": "좌측 철문·복도, 상단 대들보와 사슬, 우측 갈고리 세트, 벽의 칼 열, 도살대 — 레퍼런스 고정 요소 유지. 다만 우측 하단 피받이 양동이가 사라지고 도살대만 보이며, 칼이 놓인 작은 보조 테이블이 새로 등장. 천장등 켜짐. 창의 깨진 형상·거미줄은 레퍼런스 톤과 일치.",
      "entities": "한국인 젊은 남성 2인. 전경 남성은 어두운 데님 재킷으로 이전 스틸 인물과 의상 일치, 짧은 검은 머리. 도살대 위 인물은 이전 스틸처럼 반듯이 누워 얼굴을 위로 향한 무력 상태로 유지되며 흙빛 상의도 일치 — 고정 프롭 조항 준수. 창밖 실루엣은 핸드폰을 든 손만 보임.",
      "hard_violations": [],
      "physics": "전경 남자는 두 발로 바닥을 딛고 서서 양팔을 위로 든 자세 — 지지 정상이며 놀라 손을 드는 순간으로 자연스러움(단 손에 들고 있던 정육도가 사라짐). 누운 인물은 도살대 상판 전체가 받치고 있고 머리·어깨·다리 접촉점이 명확. 공중에 뜬 물체 없음."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 모임 지점 뒤·측면에서 방 건너 깨진 유리창을 향하고 있고, 상단 중앙에 창과 그 너머 핸드폰 플래시(손에 든 폰의 실루엣까지)가 보인다 — 지시된 조준점과 일치. 다만 전경 남자는 창을 향해 등을 보이며 두 팔을 머리 위로 완전히 들어올린 '항복' 자세로, 시선은 창이 아닌 아래/옆을 향한다. 이전 스틸의 식칼을 치켜든 동작이 사라지고 손에 아무것도 없다.",
      "built_space": "판잣집 도축장 내부: 좌측 냉장고형 문, 천장 매달린 형광등, 창 아래 벽에 칼 걸이 줄, 우측 갈고리와 체인, 도축대. 이전 스틸의 고정 요소는 대체로 유지되나 피받이 양동이가 프레임에서 사라졌고, 대신 창 아래에 원래 없던 작은 금속 테이블(칼·나무막대 놓인)이 새로 생겼다. 도축대 위 시신은 이전 스틸(테이블 뒤쪽에 놓임)과 위치가 크게 달라져 테이블 전면 중앙으로 옮겨졌고, 담요 같은 물체가 추가되었다.",
      "entities": "한국인 남성 청년 1명(전경, 이전 스틸의 짙은 데님 재킷 유지, 그러나 얼굴은 거의 보이지 않고 헤어라인이 다르게 보임), 도축대 위 누운 남성 1명(피 묻은 밝은 상의 — 이전 스틸의 인물과 대체로 일치하나 자세/방향이 완전히 재배치됨). 창 밖 촬영자는 손과 폰 실루엣만 — 규정된 배경 요소로 허용 범위. 지후의 목걸이는 프레임 내 확인 불가.",
      "hard_violations": [
       "움직일 수 없는(의식 없는) 인물의 포즈·위치가 LOCK 조항을 어기고 도축대 위에서 완전히 재배치됨 — 이전 스틸에서는 테이블 뒤쪽 가장자리에 비스듬히 놓여 있었으나 여기서는 테이블 앞쪽 중앙에 정렬되어 눕혀졌고 담요형 소품이 추가됨",
       "계약이 언급하지 않은 소품 추가(창 아래 금속 사이드 테이블과 그 위 칼·막대) 및 명시된 배경 요소 '피받이 양동이'의 삭제"
      ],
      "physics": "전경 남자는 바닥에 두 발로 서 있고 팔은 어깨 관절로 자연스럽게 들려 있어 지지는 명확하다. 누운 시신은 도축대 상판에 완전히 얹혀 있어 지지 문제 없음. 공중 부양 요소 없음. 다만 '항복' 포즈는 이전 스틸의 칼 든 동작과 물리적으로 연속되지 않는다."
     },
     {
      "label": "B",
      "direction": "카메라는 인물 뒤·측면 가슴 높이에서 방 건너 깨진 유리창을 정면으로 향하며, 상단 중앙 창 안에 플래시 광원과 폰을 든 손 실루엣이 있다 — 지시대로. 전경 남자는 창 쪽(전방)을 향해 등/측면을 보이고 아직 완전히 돌아서지 않은 상태이며, 테이블 건너 누워 있던 남자가 상체를 들어 창 쪽/카메라 쪽으로 시선을 던진다. '아직 아무도 완전히 돌아서지 않은 첫 순간'에 부합.",
      "built_space": "이전 스틸의 고정 요소가 거의 그대로 유지: 좌측 냉장고 문·매달린 고기, 천장 형광등, 창 아래 벽 칼 걸이 줄, 우측 갈고리·체인·금속 프레임, 나무 도축대, 그리고 명시된 피받이 양동이(피 담긴 금속 통)가 우측 하단 낮은 공간 기준으로 그대로 존재. 벽의 핏자국 패턴도 일치. 구조물 부품의 개수·배치에 중복이나 결손 없음.",
      "entities": "한국인 남성 청년 1명(전경, 이전 스틸과 같은 짙은 데님 워크재킷, 짧은 검은 머리, 손에 이전 스틸의 커다란 식칼을 그대로 쥠), 그리고 도축대 뒤 밝은 피 묻은 상의의 남성 1명. 인원 추가 없음. 창 밖 촬영자는 폰과 손만 보이는 배경 요소. 지후의 목걸이는 이 프레임에 해당 인물이 없어 확인 불가.",
      "hard_violations": [
       "움직일 수 없는 것으로 보이던 도축대 위 인물이 상체를 일으켜 세운 것으로 보여, LOCK 해석에 따라 포즈 변경 소지가 있음(단 해당 인물이 의식 불명임이 확정적이지 않아 경미 취급)"
      ],
      "physics": "전경 남자는 바닥에 두 발로 서고 오른손이 식칼 손잡이를 확실히 쥐고 있어 지지 명확. 반대편 남자는 양팔/손을 도축대 상판에 짚어 상체를 지탱하고 있어 공중 부유 없음. 양동이는 바닥에 놓여 있고 피는 표면 반사로 액체답게 고임. 공중에 뜬 물체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "창·플래시 배치는 맞췄으나 이전 스틸에서 LOCK된 누운 인물의 자세·위치를 통째로 재배치했고, 명시 배경요소인 피받이 양동이를 없앤 대신 요청되지 않은 사이드 테이블과 소품을 추가했으며 전경 인물의 식칼도 사라져 연속성 손실이 크다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "가슴 높이·인물 뒤측면·상단 중앙 깨진 창의 플래시라는 카메라 지시를 충족하면서 이전 스틸의 식칼·의상·벽 핏자국·피받이 양동이 등 고정 요소를 거의 그대로 보존했고, 아직 아무도 완전히 돌아서지 않은 '첫 순간'의 반응까지 담아 변경량이 가장 적다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "창·플래시 배치는 맞췄으나 이전 스틸에서 LOCK된 누운 인물의 자세·위치를 통째로 재배치했고, 명시 배경요소인 피받이 양동이를 없앤 대신 요청되지 않은 사이드 테이블과 소품을 추가했으며 전경 인물의 식칼도 사라져 연속성 손실이 크다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "가슴 높이·인물 뒤측면·상단 중앙 깨진 창의 플래시라는 카메라 지시를 충족하면서 이전 스틸의 식칼·의상·벽 핏자국·피받이 양동이 등 고정 요소를 거의 그대로 보존했고, 아직 아무도 완전히 돌아서지 않은 '첫 순간'의 반응까지 담아 변경량이 가장 적다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "카메라는 모임 지점 뒤·측면에서 방 건너 깨진 유리창을 향하고 있고, 상단 중앙에 창과 그 너머 핸드폰 플래시(손에 든 폰의 실루엣까지)가 보인다 — 지시된 조준점과 일치. 다만 전경 남자는 창을 향해 등을 보이며 두 팔을 머리 위로 완전히 들어올린 '항복' 자세로, 시선은 창이 아닌 아래/옆을 향한다. 이전 스틸의 식칼을 치켜든 동작이 사라지고 손에 아무것도 없다.",
      "built_space": "판잣집 도축장 내부: 좌측 냉장고형 문, 천장 매달린 형광등, 창 아래 벽에 칼 걸이 줄, 우측 갈고리와 체인, 도축대. 이전 스틸의 고정 요소는 대체로 유지되나 피받이 양동이가 프레임에서 사라졌고, 대신 창 아래에 원래 없던 작은 금속 테이블(칼·나무막대 놓인)이 새로 생겼다. 도축대 위 시신은 이전 스틸(테이블 뒤쪽에 놓임)과 위치가 크게 달라져 테이블 전면 중앙으로 옮겨졌고, 담요 같은 물체가 추가되었다.",
      "entities": "한국인 남성 청년 1명(전경, 이전 스틸의 짙은 데님 재킷 유지, 그러나 얼굴은 거의 보이지 않고 헤어라인이 다르게 보임), 도축대 위 누운 남성 1명(피 묻은 밝은 상의 — 이전 스틸의 인물과 대체로 일치하나 자세/방향이 완전히 재배치됨). 창 밖 촬영자는 손과 폰 실루엣만 — 규정된 배경 요소로 허용 범위. 지후의 목걸이는 프레임 내 확인 불가.",
      "hard_violations": [
       "움직일 수 없는(의식 없는) 인물의 포즈·위치가 LOCK 조항을 어기고 도축대 위에서 완전히 재배치됨 — 이전 스틸에서는 테이블 뒤쪽 가장자리에 비스듬히 놓여 있었으나 여기서는 테이블 앞쪽 중앙에 정렬되어 눕혀졌고 담요형 소품이 추가됨",
       "계약이 언급하지 않은 소품 추가(창 아래 금속 사이드 테이블과 그 위 칼·막대) 및 명시된 배경 요소 '피받이 양동이'의 삭제"
      ],
      "physics": "전경 남자는 바닥에 두 발로 서 있고 팔은 어깨 관절로 자연스럽게 들려 있어 지지는 명확하다. 누운 시신은 도축대 상판에 완전히 얹혀 있어 지지 문제 없음. 공중 부양 요소 없음. 다만 '항복' 포즈는 이전 스틸의 칼 든 동작과 물리적으로 연속되지 않는다."
     },
     {
      "label": "A",
      "direction": "카메라는 인물 뒤·측면 가슴 높이에서 방 건너 깨진 유리창을 정면으로 향하며, 상단 중앙 창 안에 플래시 광원과 폰을 든 손 실루엣이 있다 — 지시대로. 전경 남자는 창 쪽(전방)을 향해 등/측면을 보이고 아직 완전히 돌아서지 않은 상태이며, 테이블 건너 누워 있던 남자가 상체를 들어 창 쪽/카메라 쪽으로 시선을 던진다. '아직 아무도 완전히 돌아서지 않은 첫 순간'에 부합.",
      "built_space": "이전 스틸의 고정 요소가 거의 그대로 유지: 좌측 냉장고 문·매달린 고기, 천장 형광등, 창 아래 벽 칼 걸이 줄, 우측 갈고리·체인·금속 프레임, 나무 도축대, 그리고 명시된 피받이 양동이(피 담긴 금속 통)가 우측 하단 낮은 공간 기준으로 그대로 존재. 벽의 핏자국 패턴도 일치. 구조물 부품의 개수·배치에 중복이나 결손 없음.",
      "entities": "한국인 남성 청년 1명(전경, 이전 스틸과 같은 짙은 데님 워크재킷, 짧은 검은 머리, 손에 이전 스틸의 커다란 식칼을 그대로 쥠), 그리고 도축대 뒤 밝은 피 묻은 상의의 남성 1명. 인원 추가 없음. 창 밖 촬영자는 폰과 손만 보이는 배경 요소. 지후의 목걸이는 이 프레임에 해당 인물이 없어 확인 불가.",
      "hard_violations": [
       "움직일 수 없는 것으로 보이던 도축대 위 인물이 상체를 일으켜 세운 것으로 보여, LOCK 해석에 따라 포즈 변경 소지가 있음(단 해당 인물이 의식 불명임이 확정적이지 않아 경미 취급)"
      ],
      "physics": "전경 남자는 바닥에 두 발로 서고 오른손이 식칼 손잡이를 확실히 쥐고 있어 지지 명확. 반대편 남자는 양팔/손을 도축대 상판에 짚어 상체를 지탱하고 있어 공중 부유 없음. 양동이는 바닥에 놓여 있고 피는 표면 반사로 액체답게 고임. 공중에 뜬 물체 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 11
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev만 (배경 전용·공유 계획)",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S32sh10"
  },
  "lane_policy": "share_plan_prev_bgonly"
 },
 "S32sh14": {
  "input_fingerprint": "12b3a07933b0ec63",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 피를 뒤집어쓴 채 짐승처럼 사나운 눈빛으로 창밖을 노려보는 마을 청년들의 클로즈업.\n\nLOCATION (lock): The blood-spattered slaughter area immediately inside the broken window. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Complete the eye-height dolly directly along the youths' shared eyeline, ending at tight facial distance with two blood-covered faces compressed across the frame. Their fierce eyes align toward the camera because it occupies the broken-window axis, while a narrow strip of the tool-lined interior remains between their heads to preserve location and threat.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: blood-covered village youth in the middle-left of the frame, foreground, looks toward broken window on the camera axis; second blood-covered village youth in the middle-right of the frame, foreground, looks toward broken window on the camera axis; broken window on the camera axis in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 벽에 매달린 도축 도구 (Hanging along the interior wall) — Only partial hanging edges and hooked profiles are visible around the youths' heads; used as A narrow background reminder of the slaughterhouse remains visible between and beside the close faces; 깨진 유리창 (Glass broken; dust and cobwebs present) — Positioned on the camera axis opposite the youths; its broken opening leads to 김여상 and 김지후(강수리영) outside; used as Defines the off-image target axis of the youths' shared glare.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The phone-camera flash exposes dark red blood across the youths' faces in harsh, high-contrast light against the dark interior.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same slaughterhouse darkness, harsh flash illumination, bloodied tools, and workers' blood-covered state. Exclude the phone and outside observers from the tight close-up; show the youths glaring toward the window.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The village youths remain splattered with blood, and Jihu continues to wear her animal-bone necklace.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 마을 청년 #1 (Korean young man, short dark hair, broad youthful face); 마을 청년 #2 (Korean young man, short dark hair, lean youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "두 청년 모두 카메라 렌즈(=깨진 창 축)를 정면으로 응시한다. 시선이 프레임 중앙 배경의 깨진 창 축과 일치하며, 눈썹을 내리깐 사나운 표정이 카메라로 향한다. 창은 중앙 배경, 두 얼굴 사이 좁은 간격에 위치.",
    "built_space": "실내 도축장 벽면: 중앙에 창틀 하나(거미줄 낀 유리, 중앙에 깨진 구멍) 정확히 1개, 그 아래 벽걸이 칼걸이에 도축칼 여러 자루, 좌우 가장자리에 체인과 갈고리. 창 개구부는 하나이며 중복 구조물 없음. 벽·페인트 벗겨짐·핏자국 등 이전 스틸의 재질과 일치.",
    "entities": "좌측 청년: 넓적한 얼굴형·짧은 검은머리, 크림/베이지 셔츠 — 마을 청년 #1 계열로 읽힘. 우측 청년: 마른 얼굴·짧은 머리, 붉은 체크 셔츠 + 흰 이너 — 마을 청년 #2 레퍼런스 의상과 일치. 외부 관찰자(김여상·김지후)나 휴대폰은 프레임에 없음(지시대로 배제). 눈은 정상 홍채·공막.",
    "hard_violations": [],
    "physics": "두 인물 모두 서 있는 흉상 프레임으로 어깨·목이 자연스럽게 지지됨. 공중에 뜬 물체 없음. 피는 젖은 광택으로 피부 굴곡을 따라 흐르고 플래시 하이라이트가 습기에 반응 — 스티커 느낌 없음. 벽걸이 칼과 체인은 걸이·못에 매달려 무게가 아래로 향함."
   },
   {
    "label": "B",
    "direction": "두 청년 모두 카메라 쪽을 보지만, 창 개구부 안에 외부 관찰자 세 명의 얼굴이 또렷이 보인다. 시선축 자체는 카메라와 대체로 일치하나 우측 인물은 표정이 사납기보다 멍한 편.",
    "built_space": "창틀 1개, 그 아래 칼걸이와 갈고리. 구조 자체는 무리 없으나 창 너머에 세 명의 인물이 추가로 등장 — 프롬프트는 외부 관찰자를 타이트 클로즈업에서 제외하라고 명시했고, 등장 가능 인물은 마을 청년 #1·#2로 한정된다.",
    "entities": "좌측: 얼굴 전체에 짙은 붉은 분장, 회색 티셔츠 — 청년 #1로 볼 수 있으나 의상이 이전 스틸의 밝은 셔츠와 다름. 우측: 체크 셔츠로 청년 #2와 부합하나 얼굴의 붉은 분장이 거의 없어 '피를 뒤집어쓴' 상태와 어긋남. 창 안 3인은 프롬프트가 허용하지 않은 추가 인물.",
    "hard_violations": [
     "프롬프트가 명시적으로 배제한 외부 관찰자를 창 너머에 3인 렌더링 — 허용 인물 목록 밖의 인물 추가",
     "지정 인원(청년 2인)을 초과하는 추가 신체 등장"
    ],
    "physics": "두 청년은 서 있고 지지 문제 없음. 창 너머 인물들도 서 있는 것으로 읽힘. 재질감은 양호하나 우측 인물 얼굴의 분장이 거의 없어 연속성 파손."
   },
   {
    "label": "C",
    "direction": "두 청년 모두 카메라 정면을 응시하며 시선축이 창축과 일치. 다만 표정이 '짐승처럼 사나운' 강도보다는 다소 무표정/멍한 쪽. 창은 중앙 배경 상단에 위치하며 시선 목표축을 정의.",
    "built_space": "창틀 1개(깨진 유리 구멍), 그 아래 칼걸이에 도축칼 다수, 좌측 벽에 갈고리형 도구. 구조 중복 없음. 다만 창이 프레임 상단으로 밀려 두 얼굴 사이 배경 노출이 A보다 좁고 창 하단이 잘림.",
    "entities": "좌측: 넓적한 얼굴, 젖은 짧은 머리, 어두운 조끼/셔츠 — 청년 #1 계열이나 의상이 이전 스틸의 밝은 셔츠와 다름. 우측: 마른 얼굴, 짙은 회색 셔츠 — 청년 #2 레퍼런스의 붉은 체크 셔츠와 불일치. 외부 인물·휴대폰 없음. 눈은 정상 해부학.",
    "hard_violations": [],
    "physics": "두 인물 모두 서 있는 흉상, 지지 정상. 부유 물체 없음. 붉은 분장은 습윤 광택으로 피부에 얹혀 있고 플래시 반사가 자연스러움. 벽걸이 칼은 걸이에 매달려 있음."
   }
  ],
  "totals": {
   "A": 9,
   "C": 6,
   "B": 2
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 9,
    "verdict_ko": "카메라=창축 정면 응시, 두 얼굴이 좌우 전경을 채우고 중앙 배경에 깨진 창과 칼걸이가 좁게 남는 지정 레이아웃을 가장 정확히 구현했고, 외부 관찰자·휴대폰을 배제했으며 두 인물 모두 붉은 분장 상태와 의상(베이지 셔츠/붉은 체크)이 레퍼런스와 부합."
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "구도와 시선축은 맞고 추가 인물도 없으나, 우측 청년의 의상이 레퍼런스의 붉은 체크 셔츠가 아니고 창이 상단으로 밀려 배경 스트립 배치가 덜 정확하며 표정이 '사나운 눈빛'에 못 미침."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "타이트 클로즈업에서 제외하라고 명시된 외부 관찰자 3인을 창 너머에 또렷이 렌더링해 허용 인물 목록을 위반했고, 우측 청년은 붉은 분장이 거의 없어 '피를 뒤집어쓴' 지속 상태도 깨졌다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S32sh12_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 마을 청년 #1: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:976931>"
   },
   {
    "label": "CHARACTER REFERENCE — 마을 청년 #2: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1020350>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "오른쪽 청년이 빨간색 체크무늬 셔츠를 입고 있으나, 이전 샷 스틸에 고정된 대로 어두운 색의 작업용 재킷을 입어야 합니다.",
     "fix_en": "Change the right youth's clothing to a dark, heavy work jacket covered in blood splatters. Preserve his face, his expression, the left youth and his clothing, the blood on both faces, the background window, the hanging tools, the framing, and the harsh front lighting exactly.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the right youth's clothing to a dark, heavy work jacket covered in blood splatters. Preserve his face, his expression, the left youth and his clothing, the blood on both faces, the background window, the hanging tools, the framing, and the harsh front lighting exactly.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 청년 모두 얼굴이 정면 카메라를 향하고 시선도 렌즈(=깨진 창 축)로 직결된다. 왼쪽 청년은 눈썹을 내리깔고 아래턱을 당긴 채 렌즈를 노려보고, 오른쪽 청년도 동일 축으로 응시. 프레임 중앙 배경에 깨진 유리창이 있고 그 너머 바깥에서 들어오는 플래시 광점이 보여, 시선 축과 창 축이 일치한다.",
      "built_space": "도살장 내부: 좌우 벽에 사슬, 우측에 갈고리, 중앙 창 아래 벽에 칼·클리버가 걸린 도구 랙이 두 머리 사이 좁은 띠로 보인다. 창틀은 이전 스틸과 같은 격자 창에 중앙 파손 구멍, 거미줄/균열 유지. 배경 요소가 전경화되지 않고 실제 거리만큼의 화면 면적만 차지.",
      "entities": "왼쪽=넓은 얼굴형의 청년(#1 계열), 밝은 베이지 셔츠, 얼굴·목·옷에 짙은 붉은 피가 두껍게 덮임. 오른쪽=마른 얼굴의 청년(#2), 붉은 체크 셔츠+속 흰 티로 캐릭터 레퍼런스 의상과 일치, 얼굴 전면에 피 튀김. 외부 관찰자·폰 없음, 인물 수 2명으로 정확.",
      "hard_violations": [],
      "physics": "두 인물 모두 어깨·상체가 프레임 하단에 안정적으로 자리하며 서 있는 상반신. 공중에 뜬 물체 없음. 피는 젖은 광택과 흘러내린 줄기를 가진 액체로 렌더되어 표면에 얹힌 스티커가 아님. 플래시 하이라이트가 젖은 피부에 정합적으로 맺힘."
     },
     {
      "label": "B",
      "direction": "두 얼굴 모두 카메라 정면을 응시하며 창 축과 시선이 일치. 방향 자체는 맞으나 눈빛은 '짐승처럼 사나운'보다 멍하고 지친 표정에 가깝다.",
      "built_space": "A와 동일한 세트 구조: 좌측 사슬, 우측 갈고리, 중앙 창 아래 칼 랙, 격자 깨진 창. 창은 A보다 유리 파편 정리된 느낌이나 위치·비율은 동일. 배경 처리 문제 없음.",
      "entities": "왼쪽=흰 셔츠, 오른쪽=칙칙한 갈색/올리브 셔츠. 레퍼런스 #2의 붉은 체크 셔츠가 사라져 의상 일치가 무너짐. 결정적으로 '피를 뒤집어쓴' 상태가 옅은 흙먼지/주근깨 같은 얼룩으로 바뀌어 붉은 피가 거의 없음 — CARRIED STATE(피 범벅 유지)와 조명 지시(짙은 붉은 피를 플래시가 드러냄) 미충족.",
      "hard_violations": [],
      "physics": "상반신 정립, 지지 문제 없음. 다만 얼굴 얼룩이 액체 광택 없이 건조한 반점으로 찍혀 혈액의 물성(점성·흐름·젖은 반사)이 표현되지 않음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "창 축 시선·클로즈업 구도·머리 사이 도구 띠를 모두 충족하고, '피를 뒤집어쓴' 상태와 플래시가 드러내는 짙은 붉은 피, #2의 체크 셔츠까지 레퍼런스와 일치한다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "구도와 시선 축은 같지만 계약이 요구한 적 없는 변경으로 피를 흙먼지 얼룩 수준으로 지워버렸고 #2의 붉은 체크 셔츠까지 바꿔, '피를 뒤집어쓴 채 사나운 눈빛' 지시와 CARRIED STATE를 잃었다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "창 축 시선·클로즈업 구도·머리 사이 도구 띠를 모두 충족하고, '피를 뒤집어쓴' 상태와 플래시가 드러내는 짙은 붉은 피, #2의 체크 셔츠까지 레퍼런스와 일치한다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "구도와 시선 축은 같지만 계약이 요구한 적 없는 변경으로 피를 흙먼지 얼룩 수준으로 지워버렸고 #2의 붉은 체크 셔츠까지 바꿔, '피를 뒤집어쓴 채 사나운 눈빛' 지시와 CARRIED STATE를 잃었다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 청년 모두 얼굴이 정면 카메라를 향하고 시선도 렌즈(=깨진 창 축)로 직결된다. 왼쪽 청년은 눈썹을 내리깔고 아래턱을 당긴 채 렌즈를 노려보고, 오른쪽 청년도 동일 축으로 응시. 프레임 중앙 배경에 깨진 유리창이 있고 그 너머 바깥에서 들어오는 플래시 광점이 보여, 시선 축과 창 축이 일치한다.",
      "built_space": "도살장 내부: 좌우 벽에 사슬, 우측에 갈고리, 중앙 창 아래 벽에 칼·클리버가 걸린 도구 랙이 두 머리 사이 좁은 띠로 보인다. 창틀은 이전 스틸과 같은 격자 창에 중앙 파손 구멍, 거미줄/균열 유지. 배경 요소가 전경화되지 않고 실제 거리만큼의 화면 면적만 차지.",
      "entities": "왼쪽=넓은 얼굴형의 청년(#1 계열), 밝은 베이지 셔츠, 얼굴·목·옷에 짙은 붉은 피가 두껍게 덮임. 오른쪽=마른 얼굴의 청년(#2), 붉은 체크 셔츠+속 흰 티로 캐릭터 레퍼런스 의상과 일치, 얼굴 전면에 피 튀김. 외부 관찰자·폰 없음, 인물 수 2명으로 정확.",
      "hard_violations": [],
      "physics": "두 인물 모두 어깨·상체가 프레임 하단에 안정적으로 자리하며 서 있는 상반신. 공중에 뜬 물체 없음. 피는 젖은 광택과 흘러내린 줄기를 가진 액체로 렌더되어 표면에 얹힌 스티커가 아님. 플래시 하이라이트가 젖은 피부에 정합적으로 맺힘."
     },
     {
      "label": "B",
      "direction": "두 얼굴 모두 카메라 정면을 응시하며 창 축과 시선이 일치. 방향 자체는 맞으나 눈빛은 '짐승처럼 사나운'보다 멍하고 지친 표정에 가깝다.",
      "built_space": "A와 동일한 세트 구조: 좌측 사슬, 우측 갈고리, 중앙 창 아래 칼 랙, 격자 깨진 창. 창은 A보다 유리 파편 정리된 느낌이나 위치·비율은 동일. 배경 처리 문제 없음.",
      "entities": "왼쪽=흰 셔츠, 오른쪽=칙칙한 갈색/올리브 셔츠. 레퍼런스 #2의 붉은 체크 셔츠가 사라져 의상 일치가 무너짐. 결정적으로 '피를 뒤집어쓴' 상태가 옅은 흙먼지/주근깨 같은 얼룩으로 바뀌어 붉은 피가 거의 없음 — CARRIED STATE(피 범벅 유지)와 조명 지시(짙은 붉은 피를 플래시가 드러냄) 미충족.",
      "hard_violations": [],
      "physics": "상반신 정립, 지지 문제 없음. 다만 얼굴 얼룩이 액체 광택 없이 건조한 반점으로 찍혀 혈액의 물성(점성·흐름·젖은 반사)이 표현되지 않음."
     }
    ]
   },
   "reverse_raw": {
    "all_candidates_fail": false,
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "readings": [
     {
      "label": "A",
      "direction": "두 청년 모두 정면 카메라(=깨진 창 축)를 응시한다. 좌측 인물의 시선은 렌즈에 정확히 꽂히고, 우측 인물도 미세하게 좌측이지만 카메라 축을 향한다. 배경 깨진 창은 프레임 정중앙, 두 머리 사이 좁은 띠에 위치해 시선의 오프이미지 표적 축을 정의한다. 다만 눈빛이 '짐승처럼 사나운'이라기보다 멍하고 지친 표정에 가깝다.",
      "built_space": "도축장 내부: 좌우 벽에 사슬, 우측 고기갈고리, 중앙 창 아래 벽걸이 칼 세트가 걸려 있어 참조 스틸의 구조와 일치. 창은 깨진 구멍과 거미줄 균열이 있고 크기도 실제 스케일에 맞게 작게 유지됨. 구조물 중복이나 잘못된 요소 없음.",
      "entities": "한국 청년 2인. 좌측은 넓은 얼굴형(청년 #1), 우측은 마른 얼굴형(청년 #2)에 가깝다. 다만 의상: 좌측 흰 셔츠, 우측 어두운 작업복으로 청년 #2 레퍼런스의 붉은 체크셔츠가 아님. 결정적으로 '피를 뒤집어쓴' 상태가 아니라 흙먼지·검댕 얼룩 수준이며 붉은 피가 거의 없다. 캐리드 스테이트(피 범벅) 미충족.",
      "hard_violations": [],
      "physics": "두 사람 모두 지면에 서 있는 흉상 프레이밍, 어깨·목 연결 정상. 공중에 뜬 물체 없음. 벽의 칼·사슬·갈고리는 모두 못/봉/고리에 걸려 매달림이 확인됨. 무리 없음."
     },
     {
      "label": "B",
      "direction": "두 청년 모두 카메라 정면(깨진 창 축)을 노려본다. 눈썹을 낮추고 턱을 당긴, 사나운 응시가 A보다 뚜렷하다. 배경 창은 중앙, 두 머리 사이 좁은 간격에 위치해 시선 표적 축을 유지.",
      "built_space": "동일한 도축장: 좌측 사슬, 우측 갈고리·사슬, 중앙 창 아래 벽걸이 칼 세트. 창의 깨진 구멍·거미줄 균열 유지. 배경 요소가 전경으로 과대화되지 않았고 구조 중복 없음.",
      "entities": "한국 청년 2인. 좌측 넓은 얼굴(청년 #1), 우측 마른 얼굴(청년 #2). 우측 인물이 붉은 체크셔츠를 입어 청년 #2 레퍼런스 의상과 일치. 두 얼굴 모두 짙은 붉은 피가 젖은 층으로 덮여 '피를 뒤집어쓴' 상태와 CARRIED STATE를 충족. 지후의 뼈 목걸이는 프레임 밖 인물이라 해당 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 서 있는 흉상, 지지 정상. 피는 표면에 젖어 흐른 광택으로 얼굴 곡면을 따라 흐르며 스티커처럼 보이지 않음. 매달린 도구는 모두 걸림점이 보임. 부유 물체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "프레이밍·시선축·배경 구성은 맞지만 '피를 뒤집어쓴 채'라는 핵심 상태와 CARRIED STATE(피 범벅)를 거의 이행하지 못했고(먼지 얼룩 수준), 청년 #2의 체크셔츠 의상도 어긋나며 눈빛도 사납기보다 멍하다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 구도·창 중앙 배치·도축 도구 배경을 유지하면서 '피를 뒤집어쓴' 상태와 사나운 응시를 제대로 실현했고, 우측 인물의 붉은 체크셔츠로 청년 #2 레퍼런스 의상까지 맞췄다 — 추가된 변경이 모두 계약이 요구한 항목이라 손실이 없다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "프레이밍·시선축·배경 구성은 맞지만 '피를 뒤집어쓴 채'라는 핵심 상태와 CARRIED STATE(피 범벅)를 거의 이행하지 못했고(먼지 얼룩 수준), 청년 #2의 체크셔츠 의상도 어긋나며 눈빛도 사납기보다 멍하다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 구도·창 중앙 배치·도축 도구 배경을 유지하면서 '피를 뒤집어쓴' 상태와 사나운 응시를 제대로 실현했고, 우측 인물의 붉은 체크셔츠로 청년 #2 레퍼런스 의상까지 맞췄다 — 추가된 변경이 모두 계약이 요구한 항목이라 손실이 없다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "두 청년 모두 정면 카메라(=깨진 창 축)를 응시한다. 좌측 인물의 시선은 렌즈에 정확히 꽂히고, 우측 인물도 미세하게 좌측이지만 카메라 축을 향한다. 배경 깨진 창은 프레임 정중앙, 두 머리 사이 좁은 띠에 위치해 시선의 오프이미지 표적 축을 정의한다. 다만 눈빛이 '짐승처럼 사나운'이라기보다 멍하고 지친 표정에 가깝다.",
      "built_space": "도축장 내부: 좌우 벽에 사슬, 우측 고기갈고리, 중앙 창 아래 벽걸이 칼 세트가 걸려 있어 참조 스틸의 구조와 일치. 창은 깨진 구멍과 거미줄 균열이 있고 크기도 실제 스케일에 맞게 작게 유지됨. 구조물 중복이나 잘못된 요소 없음.",
      "entities": "한국 청년 2인. 좌측은 넓은 얼굴형(청년 #1), 우측은 마른 얼굴형(청년 #2)에 가깝다. 다만 의상: 좌측 흰 셔츠, 우측 어두운 작업복으로 청년 #2 레퍼런스의 붉은 체크셔츠가 아님. 결정적으로 '피를 뒤집어쓴' 상태가 아니라 흙먼지·검댕 얼룩 수준이며 붉은 피가 거의 없다. 캐리드 스테이트(피 범벅) 미충족.",
      "hard_violations": [],
      "physics": "두 사람 모두 지면에 서 있는 흉상 프레이밍, 어깨·목 연결 정상. 공중에 뜬 물체 없음. 벽의 칼·사슬·갈고리는 모두 못/봉/고리에 걸려 매달림이 확인됨. 무리 없음."
     },
     {
      "label": "A",
      "direction": "두 청년 모두 카메라 정면(깨진 창 축)을 노려본다. 눈썹을 낮추고 턱을 당긴, 사나운 응시가 A보다 뚜렷하다. 배경 창은 중앙, 두 머리 사이 좁은 간격에 위치해 시선 표적 축을 유지.",
      "built_space": "동일한 도축장: 좌측 사슬, 우측 갈고리·사슬, 중앙 창 아래 벽걸이 칼 세트. 창의 깨진 구멍·거미줄 균열 유지. 배경 요소가 전경으로 과대화되지 않았고 구조 중복 없음.",
      "entities": "한국 청년 2인. 좌측 넓은 얼굴(청년 #1), 우측 마른 얼굴(청년 #2). 우측 인물이 붉은 체크셔츠를 입어 청년 #2 레퍼런스 의상과 일치. 두 얼굴 모두 짙은 붉은 피가 젖은 층으로 덮여 '피를 뒤집어쓴' 상태와 CARRIED STATE를 충족. 지후의 뼈 목걸이는 프레임 밖 인물이라 해당 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 서 있는 흉상, 지지 정상. 피는 표면에 젖어 흐른 광택으로 얼굴 곡면을 따라 흐르며 스티커처럼 보이지 않음. 매달린 도구는 모두 걸림점이 보임. 부유 물체 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 9
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S32sh12"
  }
 },
 "S33sh10": {
  "input_fingerprint": "23ad0b7904f35ab8",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 산들의 커다란 손이 수리영의 목을 강하게 틀어쥐어 피부가 짓눌린 압박의 순간을 담은 클로즈업.\n\nLOCATION (lock): The open forest-path junction where the narrow woodland trail emerges. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At neck height beside the pair, the dolly-in ends on a close lateral view of 고산들’s hand compressing 김지후(강수리영)’s throat, with her jaw and strained shoulder retained at the frame edge for scale. The gripping hand occupies less than the central third rather than becoming an isolated insert, and the vertical line of the forearm prepares the upward track toward his face.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: forest path (nighttime path) — A narrow portion of the path recedes behind the pair; used as Appears only as restrained peripheral context around the point of contact.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key nighttime illumination emphasizes the pressure contours of hand, throat, and jaw without introducing a visible light source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same moonlit path entrance, the two characters' positions, and the man's unchanged dark clothing. Exclude the wider pursuing crowd and scenery from the close-up; focus on his hand crushing her throat.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s animal-bone necklace remains at her throat as Sandel grips her neck.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face); 고산들 (육혈귀 각성 상태) (Korean man in his mid-40s, closely cropped dark hair, cold angular face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "남자는 여자를 정면으로 응시하고, 여자도 남자를 마주본다. 그러나 남자의 손은 목을 쥐지 않고 손바닥을 펼친 채 두 사람 사이 허공에 세워져 있다 — 손끝이 위를 향하고 여자의 목·턱 어디에도 닿지 않는다. 샷 텍스트가 지정한 '목을 강하게 틀어쥔' 접촉점이 화면에 존재하지 않는다.",
    "built_space": "실내 구조물 없음. 배경은 밤의 숲길로, 두 인물 사이로 좁은 길이 후퇴하며 주변 맥락으로만 보인다. 배경 스케일은 과장 없이 적절.",
    "entities": "여자: 20대 초반 한국 여성, 긴 검은 머리, 올리브 헨리 셔츠와 갈색 자켓, 목에 구슬+동물뼈 목걸이 — 레퍼런스와 일치. 남자: 40대 중반, 짧게 친 머리, 각진 얼굴, 어두운 자켓 — 이전 스틸의 의상과 일치(단 이전 스틸 남자는 훨씬 젊어 보임). 뼈 목걸이 존재 확인.",
    "hard_violations": [],
    "physics": "손은 남자의 팔뚝-어깨로 지지되어 있으나 어떤 대상에도 접촉하지 않는다. 무게 전달·피부 압박 흔적이 전무하여 '짓눌린 압박의 순간'이라는 동작이 물리적으로 성립하지 않는다. 두 인물 모두 지면에 서 있어 부양 문제는 없음."
   },
   {
    "label": "B",
    "direction": "남자의 손이 여자의 목 아래-쇄골 상단을 움켜쥐고 있고, 손가락이 목 옆 근육을 파고들며 실제 접촉·압박이 보인다. 여자의 시선은 위쪽 남자의 얼굴로 향하고 턱이 들려 압박에 반응한다. 남자의 시선도 여자 쪽. 조준(접촉점)이 샷 텍스트의 대상에 명확히 안착.",
    "built_space": "구조물 없음. 배경 숲길이 두 인물 사이 중앙에서 좁게 후퇴하며 억제된 주변 맥락으로 기능 — 이전 스틸의 달빛 길 입구와 재질·톤이 일치.",
    "entities": "여자: 20대 초반 한국 여성, 긴 검은 생머리, 올리브 헨리 상의, 구슬줄+동물뼈 펜던트가 목/가슴에 그대로 있음 — 레퍼런스 일치. 다만 겉의 갈색 자켓이 프레임 안에서 보이지 않는다(어깨 크롭·어둠으로 잘림 가능). 남자: 40대 중반, 짧은 머리, 각진 옆얼굴, 어두운 자켓 — 일치.",
    "hard_violations": [],
    "physics": "손은 남자의 뻗은 팔로 지지되고, 손가락 관절이 힘줄로 긴장되어 있으며 여자의 옷·피부가 눌려 접히는 반응이 보인다. 여자는 지면에 서서 목이 뒤로 밀리는 자세 — 압박 동작과 일관. 부양 물체 없음."
   },
   {
    "label": "C",
    "direction": "남자의 손은 여자의 목이 아니라 자켓 깃/옷섶을 쥐고 있다. 손가락이 천을 잡고 있어 목 피부 압박이 아니라 멱살잡이에 가깝다. 남자의 시선은 여자 얼굴로, 여자의 시선은 프레임 왼쪽 아래(남자 쪽 방향이나 초점은 어긋남). 접촉 대상이 '목'이 아니라 '옷'이라는 점에서 샷 텍스트 조준이 부분적으로 빗나감.",
    "built_space": "구조물 없음. 배경은 어두운 숲과 오른쪽으로 살짝 보이는 길 — 이전 스틸의 길 입구 느낌은 약하지만 장소 톤은 유지.",
    "entities": "여자: 20대 초반 한국 여성, 긴 검은 머리, 올리브 헨리+갈색 자켓, 구슬+동물뼈 목걸이 — 레퍼런스 일치도 높음. 남자: 40대 중반, 짧은 머리, 각진 얼굴, 어두운 자켓 — 일치.",
    "hard_violations": [],
    "physics": "손은 팔로 지지되고 실제 천을 쥐어 옷 주름이 손아귀 방향으로 당겨져 있다 — 물리적으로 성립. 다만 목 피부의 눌림 변형은 전무. 두 인물 모두 지면에 서 있음."
   }
  ],
  "totals": {
   "A": 3,
   "B": 8,
   "C": 5
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "손이 목에 전혀 닿지 않고 허공에 펼쳐져 있어 '목을 강하게 틀어쥔 압박의 순간'이라는 핵심 동작 자체가 부재하며, 프레이밍도 두 얼굴을 다 담은 미디엄에 가까워 클로즈업 지시에서 가장 멀다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "손가락이 목을 실제로 파고들어 압박 접촉이 유일하게 성립하고, 목높이 측면 근접 프레임에 턱·긴장된 어깨가 가장자리에 남아 샷 텍스트의 구성과 뼈 목걸이 지속 조건을 모두 충족한다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "장소·의상·목걸이 재현은 좋으나 손이 목이 아닌 옷깃을 쥐어 '피부가 짓눌린 압박' 순간을 놓쳤고, 프레임도 클로즈업보다 넓다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S33sh6_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:813104>"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들 (육혈귀 각성 상태): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:990262>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "남자의 손이 여자의 맨 목이 아닌 셔츠 옷깃을 잡고 있어, 목 피부가 짓눌리는 샷 텍스트의 핵심 액션을 위반했습니다.",
     "fix_en": "Redraw the man's hand and the woman's neck: his bare hand must directly wrap around and tightly grip her bare throat, with his fingers visibly pressing into her skin to show physical compression. Remove the grip on the fabric, letting her shirt collar rest normally below his hand. Strictly preserve the characters' faces, the bone necklace, the man's clothing, the forest background, and the exact framing.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "이전 샷에서 설정된 여자의 짙은 갈색 재킷이 사라지고 올리브색 이너 셔츠만 보입니다.",
     "fix_en": "Draw the dark brown jacket over the woman's visible shoulder and back, matching the previous shot still. The jacket must be worn over her current olive green shirt. Keep the man, his hand position, her face and hair, the bone necklace, the background, and the lighting exactly as they are.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the man's hand and the woman's neck: his bare hand must directly wrap around and tightly grip her bare throat, with his fingers visibly pressing into her skin to show physical compression. Remove the grip on the fabric, letting her shirt collar rest normally below his hand. Strictly preserve the characters' faces, the bone necklace, the man's clothing, the forest background, and the exact framing.\n- Draw the dark brown jacket over the woman's visible shoulder and back, matching the previous shot still. The jacket must be worn over her current olive green shirt. Keep the man, his hand position, her face and hair, the bone necklace, the background, and the lighting exactly as they are.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "all_candidates_fail": false,
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "readings": [
     {
      "label": "A",
      "direction": "남자는 프레임 좌측에서 우측의 여자를 향해 옆얼굴로 서 있고 시선은 여자 쪽을 향한다. 여자의 시선은 위쪽·남자 쪽을 올려다본다. 그러나 남자의 손은 여자의 '목'이 아니라 쇄골/어깨 앞쪽(가슴 위 옷깃 부근)을 주먹처럼 움켜쥐고 있어, 샷 텍스트가 명시한 '목을 강하게 틀어쥐어 피부가 짓눌리는' 지점에 접촉이 닿지 않는다. 카메라는 목 높이 측면 근접으로 맞다.",
      "built_space": "실내 구조물 없음. 배경은 야간 숲길 입구가 두 사람 사이 중앙으로 좁게 후퇴하며, 참조 스틸의 낙엽 깔린 길·덤불 재질과 일치. 배경 요소가 전경으로 확대되지 않았다.",
      "entities": "남자: 40대 한국 남성, 짧은 검은 머리, 각진 냉정한 얼굴, 이전 스틸의 어두운 재킷 유지 — 일치. 여자: 20대 초 긴 검은 생머리, 앳된 얼굴 — 일치. 다만 상의가 이전 스틸/캐릭터 레퍼런스의 어두운 겉옷(재킷) 대신 올리브 헨리 이너만 보여 겉옷이 사라졌다. 동물 뼈 목걸이는 목이 아니라 가슴 아래로 길게 늘어져 보인다.",
      "hard_violations": [],
      "physics": "두 사람 모두 지면에 선 자세, 남자의 팔은 어깨에서 전완으로 이어져 손이 여자 옷을 쥐고 있어 지지는 성립. 다만 압박 접촉면이 쇄골 위 천이라 '피부가 짓눌리는' 물리적 결과(목 살의 눌림, 턱 들림)가 나타나지 않는다."
     },
     {
      "label": "B",
      "direction": "동일한 좌-남/우-여 측면 근접. 남자의 손이 여자의 목(후두 아래 앞목)에 직접 얹혀 손가락이 목 피부를 누르고, 여자는 턱을 살짝 들고 위를 응시한다 — 샷 텍스트가 지정한 접촉 지점에 정확히 닿는다. 전완의 수직/사선 선이 그의 얼굴 쪽 상향 트랙을 준비한다.",
      "built_space": "구조물 없음. 배경 숲길이 중앙으로 좁게 후퇴, 참조 스틸의 재질·야간 저조도 유지. 배경 확대 없음.",
      "entities": "남자: 40대, 짧은 머리, 각진 얼굴, 어두운 재킷 — 이전 스틸과 일치. 여자: 20대 초, 긴 검은 머리 — 일치하고 캐릭터 레퍼런스의 갈색 워크 재킷 + 올리브 이너 조합을 복원. 동물 뼈 목걸이가 목 아래 가슴에 걸려 유지됨.",
      "hard_violations": [],
      "physics": "두 인물 모두 지면에 서 있고, 남자의 팔은 어깨-팔꿈치-손목으로 연결되어 손바닥이 목에 닿아 있다. 목을 누르는 손가락과 들린 턱, 긴장된 어깨선이 압박 동작과 일관된다. 다만 손아귀 힘(손가락 파고듦)은 다소 절제되어 있다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "손이 목이 아닌 쇄골·옷깃을 쥐어 '목을 틀어쥐어 피부가 짓눌리는' 핵심 접촉을 놓쳤고, 지후의 겉옷 재킷도 레퍼런스에서 사라져 의상 락을 이탈했다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "손이 실제 목 앞면을 눌러 샷 텍스트의 압박 순간을 정확히 성립시키고, 캐릭터 레퍼런스의 갈색 재킷과 뼈 목걸이·야간 숲길 배경까지 유지했다 — 얻은 것이 비용보다 크다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "손이 목이 아닌 쇄골·옷깃을 쥐어 '목을 틀어쥐어 피부가 짓눌리는' 핵심 접촉을 놓쳤고, 지후의 겉옷 재킷도 레퍼런스에서 사라져 의상 락을 이탈했다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "손이 실제 목 앞면을 눌러 샷 텍스트의 압박 순간을 정확히 성립시키고, 캐릭터 레퍼런스의 갈색 재킷과 뼈 목걸이·야간 숲길 배경까지 유지했다 — 얻은 것이 비용보다 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "남자는 프레임 좌측에서 우측의 여자를 향해 옆얼굴로 서 있고 시선은 여자 쪽을 향한다. 여자의 시선은 위쪽·남자 쪽을 올려다본다. 그러나 남자의 손은 여자의 '목'이 아니라 쇄골/어깨 앞쪽(가슴 위 옷깃 부근)을 주먹처럼 움켜쥐고 있어, 샷 텍스트가 명시한 '목을 강하게 틀어쥐어 피부가 짓눌리는' 지점에 접촉이 닿지 않는다. 카메라는 목 높이 측면 근접으로 맞다.",
      "built_space": "실내 구조물 없음. 배경은 야간 숲길 입구가 두 사람 사이 중앙으로 좁게 후퇴하며, 참조 스틸의 낙엽 깔린 길·덤불 재질과 일치. 배경 요소가 전경으로 확대되지 않았다.",
      "entities": "남자: 40대 한국 남성, 짧은 검은 머리, 각진 냉정한 얼굴, 이전 스틸의 어두운 재킷 유지 — 일치. 여자: 20대 초 긴 검은 생머리, 앳된 얼굴 — 일치. 다만 상의가 이전 스틸/캐릭터 레퍼런스의 어두운 겉옷(재킷) 대신 올리브 헨리 이너만 보여 겉옷이 사라졌다. 동물 뼈 목걸이는 목이 아니라 가슴 아래로 길게 늘어져 보인다.",
      "hard_violations": [],
      "physics": "두 사람 모두 지면에 선 자세, 남자의 팔은 어깨에서 전완으로 이어져 손이 여자 옷을 쥐고 있어 지지는 성립. 다만 압박 접촉면이 쇄골 위 천이라 '피부가 짓눌리는' 물리적 결과(목 살의 눌림, 턱 들림)가 나타나지 않는다."
     },
     {
      "label": "B",
      "direction": "동일한 좌-남/우-여 측면 근접. 남자의 손이 여자의 목(후두 아래 앞목)에 직접 얹혀 손가락이 목 피부를 누르고, 여자는 턱을 살짝 들고 위를 응시한다 — 샷 텍스트가 지정한 접촉 지점에 정확히 닿는다. 전완의 수직/사선 선이 그의 얼굴 쪽 상향 트랙을 준비한다.",
      "built_space": "구조물 없음. 배경 숲길이 중앙으로 좁게 후퇴, 참조 스틸의 재질·야간 저조도 유지. 배경 확대 없음.",
      "entities": "남자: 40대, 짧은 머리, 각진 얼굴, 어두운 재킷 — 이전 스틸과 일치. 여자: 20대 초, 긴 검은 머리 — 일치하고 캐릭터 레퍼런스의 갈색 워크 재킷 + 올리브 이너 조합을 복원. 동물 뼈 목걸이가 목 아래 가슴에 걸려 유지됨.",
      "hard_violations": [],
      "physics": "두 인물 모두 지면에 서 있고, 남자의 팔은 어깨-팔꿈치-손목으로 연결되어 손바닥이 목에 닿아 있다. 목을 누르는 손가락과 들린 턱, 긴장된 어깨선이 압박 동작과 일관된다. 다만 손아귀 힘(손가락 파고듦)은 다소 절제되어 있다."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 목 높이에서 두 사람 옆을 잡은 측면 클로즈업. 남자(고산들)의 시선은 오른쪽 여자의 얼굴을 향하고, 그의 오른팔은 화면 왼쪽에서 오른쪽으로 수평으로 뻗어 손바닥·손가락이 지후의 목 앞·쇄골 위 인후부에 닿아 있다. 지후의 시선은 위쪽 남자 얼굴 쪽으로 올라가 있다. 손의 조준점은 '목'으로 정확히 착지하나, 손가락이 목을 완전히 감싸 쥐기보다 인후를 밀어 누르는 형태.",
      "built_space": "인공 구조물 없음. 배경은 밤의 숲길 입구가 두 사람 사이 중앙에 좁게 후퇴하며 보이고, 양옆은 어두운 수풀. 이전 스틸의 낙엽 깔린 길·나무 벽이 일치. 배경 요소가 전경으로 확대되지 않음.",
      "entities": "남자: 40대 중반, 짧게 친 검은 머리, 각진 냉랭한 얼굴, 이전 스틸과 동일한 어두운 재킷 — 일치. 여자: 20대 초반 한국 여성, 긴 검은 생머리, 캐릭터 레퍼런스의 갈색 워크재킷 + 올리브 헨리 이너 착용, 목에는 나무구슬 줄에 매달린 흰 뼈(뿔) 펜던트가 그대로 보임 — 캐리드 스테이트 충족. 추가 인물 없음, 텍스트/워터마크 없음.",
      "hard_violations": [],
      "physics": "두 사람 모두 서 있고 지면에 지지됨. 남자의 팔은 어깨에서 뻗어 손이 여자 목에 접촉해 하중과 압박 방향이 일관되며, 여자의 어깨가 약간 올라가고 턱이 들려 압박 반응이 읽힌다. 공중에 떠 있는 물체·신체 없음. 다만 손가락 관절의 힘줄 긴장은 다소 약해 '강하게 틀어쥐는' 강도는 중간 수준."
     },
     {
      "label": "B",
      "direction": "동일한 측면 클로즈업, 남자 시선은 여자 얼굴로. 그러나 그의 주먹 쥔 손은 여자의 목이 아니라 어깨 안쪽/쇄골 상단의 옷깃 부근을 움켜쥐고 있다. 목(인후)은 손 위쪽으로 완전히 노출되어 접촉 없음 — 샷 텍스트가 지정한 '목을 틀어쥔' 착지점을 빗나감.",
      "built_space": "동일한 밤 숲길 입구, 배경 일치. 구조물 없음.",
      "entities": "남자 동일. 여자는 갈색 워크재킷이 사라지고 올리브 헨리 상의만 입고 있음 — 레퍼런스 및 이전 스틸의 겉옷(어두운 상의)에서 이탈한, 계약이 요구하지 않은 의상 변경. 뼈 목걸이는 유지. 인물 수는 정상.",
      "hard_violations": [],
      "physics": "둘 다 지면에 서 있고 팔은 어깨에서 이어져 손이 여자 옷을 잡고 있어 지지는 성립. 다만 주먹이 옷 천을 잡은 형태라 '피부가 짓눌리는 압박'의 물리적 결과(목 살의 눌림)가 나타나지 않는다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "손이 실제로 인후에 닿아 '목을 틀어쥔 압박의 순간'을 담았고, 이전 스틸·캐릭터 레퍼런스의 겉옷과 뼈 목걸이가 모두 유지되어 계약 이탈이 없다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "손이 목이 아닌 쇄골·옷깃을 쥐어 샷 텍스트의 핵심 접촉점을 빗나갔고, 게다가 요구되지 않은 겉옷 제거로 지후의 의상까지 변경해 얻은 것보다 잃은 것이 크다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "손이 실제로 인후에 닿아 '목을 틀어쥔 압박의 순간'을 담았고, 이전 스틸·캐릭터 레퍼런스의 겉옷과 뼈 목걸이가 모두 유지되어 계약 이탈이 없다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "손이 목이 아닌 쇄골·옷깃을 쥐어 샷 텍스트의 핵심 접촉점을 빗나갔고, 게다가 요구되지 않은 겉옷 제거로 지후의 의상까지 변경해 얻은 것보다 잃은 것이 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "카메라는 목 높이에서 두 사람 옆을 잡은 측면 클로즈업. 남자(고산들)의 시선은 오른쪽 여자의 얼굴을 향하고, 그의 오른팔은 화면 왼쪽에서 오른쪽으로 수평으로 뻗어 손바닥·손가락이 지후의 목 앞·쇄골 위 인후부에 닿아 있다. 지후의 시선은 위쪽 남자 얼굴 쪽으로 올라가 있다. 손의 조준점은 '목'으로 정확히 착지하나, 손가락이 목을 완전히 감싸 쥐기보다 인후를 밀어 누르는 형태.",
      "built_space": "인공 구조물 없음. 배경은 밤의 숲길 입구가 두 사람 사이 중앙에 좁게 후퇴하며 보이고, 양옆은 어두운 수풀. 이전 스틸의 낙엽 깔린 길·나무 벽이 일치. 배경 요소가 전경으로 확대되지 않음.",
      "entities": "남자: 40대 중반, 짧게 친 검은 머리, 각진 냉랭한 얼굴, 이전 스틸과 동일한 어두운 재킷 — 일치. 여자: 20대 초반 한국 여성, 긴 검은 생머리, 캐릭터 레퍼런스의 갈색 워크재킷 + 올리브 헨리 이너 착용, 목에는 나무구슬 줄에 매달린 흰 뼈(뿔) 펜던트가 그대로 보임 — 캐리드 스테이트 충족. 추가 인물 없음, 텍스트/워터마크 없음.",
      "hard_violations": [],
      "physics": "두 사람 모두 서 있고 지면에 지지됨. 남자의 팔은 어깨에서 뻗어 손이 여자 목에 접촉해 하중과 압박 방향이 일관되며, 여자의 어깨가 약간 올라가고 턱이 들려 압박 반응이 읽힌다. 공중에 떠 있는 물체·신체 없음. 다만 손가락 관절의 힘줄 긴장은 다소 약해 '강하게 틀어쥐는' 강도는 중간 수준."
     },
     {
      "label": "A",
      "direction": "동일한 측면 클로즈업, 남자 시선은 여자 얼굴로. 그러나 그의 주먹 쥔 손은 여자의 목이 아니라 어깨 안쪽/쇄골 상단의 옷깃 부근을 움켜쥐고 있다. 목(인후)은 손 위쪽으로 완전히 노출되어 접촉 없음 — 샷 텍스트가 지정한 '목을 틀어쥔' 착지점을 빗나감.",
      "built_space": "동일한 밤 숲길 입구, 배경 일치. 구조물 없음.",
      "entities": "남자 동일. 여자는 갈색 워크재킷이 사라지고 올리브 헨리 상의만 입고 있음 — 레퍼런스 및 이전 스틸의 겉옷(어두운 상의)에서 이탈한, 계약이 요구하지 않은 의상 변경. 뼈 목걸이는 유지. 인물 수는 정상.",
      "hard_violations": [],
      "physics": "둘 다 지면에 서 있고 팔은 어깨에서 이어져 손이 여자 옷을 잡고 있어 지지는 성립. 다만 주먹이 옷 천을 잡은 형태라 '피부가 짓눌리는 압박'의 물리적 결과(목 살의 눌림)가 나타나지 않는다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 9,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S33sh6"
  }
 },
 "groupbg::꿈속 오솔길": {
  "input_fingerprint": "639c3e079dcefbde",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "꿈속 오솔길",
    "tags": [
     "S35sh1",
     "S35sh6",
     "S35sh7"
    ]
   },
   "context_sig": "8eb9aa657577feb2"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated: A dreamlike forest footpath on the island, with the road stretching unnaturally between daughter and mother.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n꿈속 금월도 숲속 오솔길: 현실의 해송 숲길과 같은 배경이나, 시각적 원근감이 뒤틀린 공간. 길의 앞뒤가 비현실적으로 길어 보이며 형태가 흐릿해지는 초현실적인 톤의 좁은 흙길이다. (특징: 원근감이 왜곡된 길게 뻗은 흙길; 양옆의 빽빽한 나무 질감; 흐릿하고 비현실적인 공간감)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 숲속 오솔길 - 낮(수리영의 꿈)\n- 수리영이 오솔길에 홀로 서 있다.\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated: A dreamlike forest footpath on the island, with the road stretching unnaturally between daughter and mother.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n꿈속 금월도 숲속 오솔길: 현실의 해송 숲길과 같은 배경이나, 시각적 원근감이 뒤틀린 공간. 길의 앞뒤가 비현실적으로 길어 보이며 형태가 흐릿해지는 초현실적인 톤의 좁은 흙길이다. (특징: 원근감이 왜곡된 길게 뻗은 흙길; 양옆의 빽빽한 나무 질감; 흐릿하고 비현실적인 공간감)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 숲속 오솔길 - 낮(수리영의 꿈)\n- 수리영이 오솔길에 홀로 서 있다.\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_꿈속_오솔길_464fe0.png",
  "asset_id": "575a0972-407e-47e4-ad05-b932d41fafa4",
  "input_asset_ids": [
   "0f3823d6-8dc2-4b74-a54a-24473246f20e"
  ],
  "origin_tag": "S35sh1",
  "place_text": "A dreamlike forest footpath on the island, with the road stretching unnaturally between daughter and mother.",
  "origin_inputs": {
   "place_text": "A dreamlike forest footpath on the island, with the road stretching unnaturally between daughter and mother.",
   "time_of_day_en": "day",
   "conti_asset_id": "0f3823d6-8dc2-4b74-a54a-24473246f20e"
  }
 },
 "S35sh1::bgfirst_bg": {
  "input_fingerprint": "6017d398499437cc",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어스름한 오솔길 한가운데 홀로 서 있는 수리영의 뒷모습.\n\nLOCATION (lock): A dreamlike forest footpath on the island, with the road stretching unnaturally between daughter and mother.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Centered at shoulder height behind 김지후(강수리영), the static wide frame places her back in the lower center and 강민숙(홍은미) far ahead on the upper path. The long central sightline makes their separation visually explicit, with the path connecting them while withholding any sense that the distance can close.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the lower-center of the frame, foreground, looks toward 강민숙(홍은미) on the upper path; 강민숙(홍은미) in the upper-center of the frame, background, looks toward 김지후(강수리영) down the path; path separating mother and daughter in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: forest path (daytime path within the dream) — The path rises through the center from 김지후(강수리영) toward 강민숙(홍은미); used as Creates the impossible depth corridor linking but separating mother and daughter; forest (surrounding the path); used as Borders the long sightline and contains both figures within the same dream space.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Daylight appropriate to the dream setting is rendered with restrained contrast and subtly unreal spatial separation rather than an invented light source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어스름한 오솔길 한가운데 홀로 서 있는 수리영의 뒷모습.\n\nLOCATION (lock): A dreamlike forest footpath on the island, with the road stretching unnaturally between daughter and mother.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Centered at shoulder height behind 김지후(강수리영), the static wide frame places her back in the lower center and 강민숙(홍은미) far ahead on the upper path. The long central sightline makes their separation visually explicit, with the path connecting them while withholding any sense that the distance can close.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the lower-center of the frame, foreground, looks toward 강민숙(홍은미) on the upper path; 강민숙(홍은미) in the upper-center of the frame, background, looks toward 김지후(강수리영) down the path; path separating mother and daughter in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: forest path (daytime path within the dream) — The path rises through the center from 김지후(강수리영) toward 강민숙(홍은미); used as Creates the impossible depth corridor linking but separating mother and daughter; forest (surrounding the path); used as Borders the long sightline and contains both figures within the same dream space.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Daylight appropriate to the dream setting is rendered with restrained contrast and subtly unreal spatial separation rather than an invented light source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S35sh1__bgfirst_bg.png",
  "asset_id": "e18c9255-c6ce-4aef-8ed9-2f6097998213",
  "input_asset_ids": [
   "0f3823d6-8dc2-4b74-a54a-24473246f20e",
   "575a0972-407e-47e4-ad05-b932d41fafa4"
  ]
 },
 "S35sh1": {
  "input_fingerprint": "c263e764dc226c9a",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어스름한 오솔길 한가운데 홀로 서 있는 수리영의 뒷모습.\n\nLOCATION (lock): A dreamlike forest footpath on the island, with the road stretching unnaturally between daughter and mother. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Centered at shoulder height behind 김지후(강수리영), the static wide frame places her back in the lower center and 강민숙(홍은미) far ahead on the upper path. The long central sightline makes their separation visually explicit, with the path connecting them while withholding any sense that the distance can close.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the lower-center of the frame, foreground, looks toward 강민숙(홍은미) on the upper path; 강민숙(홍은미) in the upper-center of the frame, background, looks toward 김지후(강수리영) down the path; path separating mother and daughter in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: forest path (daytime path within the dream) — The path rises through the center from 김지후(강수리영) toward 강민숙(홍은미); used as Creates the impossible depth corridor linking but separating mother and daughter; forest (surrounding the path); used as Borders the long sightline and contains both figures within the same dream space.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Daylight appropriate to the dream setting is rendered with restrained contrast and subtly unreal spatial separation rather than an invented light source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어스름한 오솔길 한가운데 홀로 서 있는 수리영의 뒷모습.\n\nLOCATION (lock): A dreamlike forest footpath on the island, with the road stretching unnaturally between daughter and mother. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Centered at shoulder height behind 김지후(강수리영), the static wide frame places her back in the lower center and 강민숙(홍은미) far ahead on the upper path. The long central sightline makes their separation visually explicit, with the path connecting them while withholding any sense that the distance can close.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the lower-center of the frame, foreground, looks toward 강민숙(홍은미) on the upper path; 강민숙(홍은미) in the upper-center of the frame, background, looks toward 김지후(강수리영) down the path; path separating mother and daughter in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: forest path (daytime path within the dream) — The path rises through the center from 김지후(강수리영) toward 강민숙(홍은미); used as Creates the impossible depth corridor linking but separating mother and daughter; forest (surrounding the path); used as Borders the long sightline and contains both figures within the same dream space.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Daylight appropriate to the dream setting is rendered with restrained contrast and subtly unreal spatial separation rather than an invented light source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 어스름한 오솔길 한가운데 홀로 서 있는 수리영의 뒷모습.\n\nLOCATION (lock): A dreamlike forest footpath on the island, with the road stretching unnaturally between daughter and mother. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Centered at shoulder height behind 김지후(강수리영), the static wide frame places her back in the lower center and 강민숙(홍은미) far ahead on the upper path. The long central sightline makes their separation visually explicit, with the path connecting them while withholding any sense that the distance can close.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 김지후(강수리영) in the lower-center of the frame, foreground, looks toward 강민숙(홍은미) on the upper path; 강민숙(홍은미) in the upper-center of the frame, background, looks toward 김지후(강수리영) down the path; path separating mother and daughter in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: forest path (daytime path within the dream) — The path rises through the center from 김지후(강수리영) toward 강민숙(홍은미); used as Creates the impossible depth corridor linking but separating mother and daughter; forest (surrounding the path); used as Borders the long sightline and contains both figures within the same dream space.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Daylight appropriate to the dream setting is rendered with restrained contrast and subtly unreal spatial separation rather than an invented light source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S35sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S35sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:813104>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_꿈속_오솔길_464fe0.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:813104>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "전경 김지후를 하단 중앙에 작게 두고 강민숙을 상단 중앙의 먼 길에 배치해 권위적인 와이드 프레이밍과 불가능할 만큼 긴 이격감을 가장 충실히 구현했으나, 김지후의 참고 복장 일치는 약하다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "상호 시선과 숲길 장소, 김지후의 재킷은 잘 맞지만 전경 인물이 지나치게 크고 왼쪽에 치우쳤으며 강민숙도 충분히 멀리 있는 상단 경로에 놓이지 않아 핵심 와이드 구도가 약하다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.5
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.5
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "전경 김지후를 하단 중앙에 작게 두고 강민숙을 상단 중앙의 먼 길에 배치해 권위적인 와이드 프레이밍과 불가능할 만큼 긴 이격감을 가장 충실히 구현했으나, 김지후의 참고 복장 일치는 약하다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "상호 시선과 숲길 장소, 김지후의 재킷은 잘 맞지만 전경 인물이 지나치게 크고 왼쪽에 치우쳤으며 강민숙도 충분히 멀리 있는 상단 경로에 놓이지 않아 핵심 와이드 구도가 약하다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 김지후의 등 뒤 어깨높이에 위치, 시선축이 오솔길 중앙을 따라 상단의 여성(강민숙)으로 곧게 이어진다. 전경 인물은 뒷모습으로 길 위쪽을 향해 서 있고, 배경 인물은 정면으로 카메라(=딸) 쪽을 바라본다. 두 시선이 같은 중앙 축에서 맞물린다.",
      "built_space": "인공 구조물은 없음. 레퍼런스의 소나무 숲 흙길과 좌측 큰 나무둥치, 좌측 하단 바위, 양옆 관목대가 그대로 재현되어 장소 락이 일치한다. 길은 중앙에서 원근으로 좁아지며 안개로 소실된다.",
      "entities": "전경 인물: 긴 검은 생머리, 캐릭터 레퍼런스의 갈색 워크재킷과 짙은 카고팬츠를 착용한 한국 여성 뒷모습 — 의상 일치(모자는 없음). 배경 인물: 중년 여성으로 보이는 한국인, 어두운 상의와 바지, 화면상 아주 작게 실제 거리 비례로 렌더됨. 문자·워터마크 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 두 발이 흙길 지면에 닿아 있고 그림자·발밑 접지가 자연스럽다. 전경 인물은 팔을 자연스럽게 늘어뜨린 채 무게중심이 안정적이며, 배경 인물도 길 위에 서 있다. 공중에 뜬 물체나 지지 없는 신체 없음."
     },
     {
      "label": "B",
      "direction": "카메라는 등 뒤 중앙축이나 높이가 다소 낮아(허리~가슴) 인물 머리가 프레임 하단 중앙에 크게 걸린다. 전경 인물은 길 위쪽을 향하고, 배경 인물은 정면으로 이쪽을 향해 서 있어 시선축 자체는 성립한다.",
      "built_space": "구조물 없음. 소나무 숲과 흙길, 좌측 바위 등 장소는 레퍼런스와 대체로 일치하나 길이 다소 좁고 관목/지면 톤이 더 회색으로 어두워 '낮'의 느낌이 약하다.",
      "entities": "전경 인물: 긴 검은 머리이나 상의가 단색 검정 니트/코트로 캐릭터 레퍼런스의 갈색 워크재킷·카고팬츠와 불일치. 배경 인물: 어두운 원피스 차림 여성으로 실루엣에 가깝고 얼굴 디테일이 뭉개져 '구체적 인간'요구를 약하게 충족. 문자 없음.",
      "hard_violations": [],
      "physics": "배경 인물은 길 위에 서 있고 발이 지면에 접한다. 전경 인물은 어깨 위만 보이며 상체가 프레임에 잘려 지지 판단은 제한적이나 부유·왜곡은 없다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "어깨높이 뒷모습·하단 중앙 배치, 상단 중앙의 어머니, 중앙 원근 통로가 모두 지시대로이며 의상·장소 락도 레퍼런스와 일치해 가장 충실하다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "구도 축은 맞으나 카메라가 어깨높이보다 낮아 전경 인물이 과하게 크게 잘리고, 검정 단색 의상이 캐릭터 레퍼런스(갈색 재킷·카고팬츠)와 어긋나며 배경 인물이 실루엣에 가깝다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "어깨높이 뒷모습·하단 중앙 배치, 상단 중앙의 어머니, 중앙 원근 통로가 모두 지시대로이며 의상·장소 락도 레퍼런스와 일치해 가장 충실하다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "구도 축은 맞으나 카메라가 어깨높이보다 낮아 전경 인물이 과하게 크게 잘리고, 검정 단색 의상이 캐릭터 레퍼런스(갈색 재킷·카고팬츠)와 어긋나며 배경 인물이 실루엣에 가깝다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "카메라는 김지후의 등 뒤 어깨높이에 위치, 시선축이 오솔길 중앙을 따라 상단의 여성(강민숙)으로 곧게 이어진다. 전경 인물은 뒷모습으로 길 위쪽을 향해 서 있고, 배경 인물은 정면으로 카메라(=딸) 쪽을 바라본다. 두 시선이 같은 중앙 축에서 맞물린다.",
      "built_space": "인공 구조물은 없음. 레퍼런스의 소나무 숲 흙길과 좌측 큰 나무둥치, 좌측 하단 바위, 양옆 관목대가 그대로 재현되어 장소 락이 일치한다. 길은 중앙에서 원근으로 좁아지며 안개로 소실된다.",
      "entities": "전경 인물: 긴 검은 생머리, 캐릭터 레퍼런스의 갈색 워크재킷과 짙은 카고팬츠를 착용한 한국 여성 뒷모습 — 의상 일치(모자는 없음). 배경 인물: 중년 여성으로 보이는 한국인, 어두운 상의와 바지, 화면상 아주 작게 실제 거리 비례로 렌더됨. 문자·워터마크 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 두 발이 흙길 지면에 닿아 있고 그림자·발밑 접지가 자연스럽다. 전경 인물은 팔을 자연스럽게 늘어뜨린 채 무게중심이 안정적이며, 배경 인물도 길 위에 서 있다. 공중에 뜬 물체나 지지 없는 신체 없음."
     },
     {
      "label": "A",
      "direction": "카메라는 등 뒤 중앙축이나 높이가 다소 낮아(허리~가슴) 인물 머리가 프레임 하단 중앙에 크게 걸린다. 전경 인물은 길 위쪽을 향하고, 배경 인물은 정면으로 이쪽을 향해 서 있어 시선축 자체는 성립한다.",
      "built_space": "구조물 없음. 소나무 숲과 흙길, 좌측 바위 등 장소는 레퍼런스와 대체로 일치하나 길이 다소 좁고 관목/지면 톤이 더 회색으로 어두워 '낮'의 느낌이 약하다.",
      "entities": "전경 인물: 긴 검은 머리이나 상의가 단색 검정 니트/코트로 캐릭터 레퍼런스의 갈색 워크재킷·카고팬츠와 불일치. 배경 인물: 어두운 원피스 차림 여성으로 실루엣에 가깝고 얼굴 디테일이 뭉개져 '구체적 인간'요구를 약하게 충족. 문자 없음.",
      "hard_violations": [],
      "physics": "배경 인물은 길 위에 서 있고 발이 지면에 접한다. 전경 인물은 어깨 위만 보이며 상체가 프레임에 잘려 지지 판단은 제한적이나 부유·왜곡은 없다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 13,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   }
  },
  "totals": {
   "A": 13,
   "B": 14
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "전경 김지후를 하단 중앙에 작게 두고 강민숙을 상단 중앙의 먼 길에 배치해 권위적인 와이드 프레이밍과 불가능할 만큼 긴 이격감을 가장 충실히 구현했으나, 김지후의 참고 복장 일치는 약하다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "상호 시선과 숲길 장소, 김지후의 재킷은 잘 맞지만 전경 인물이 지나치게 크고 왼쪽에 치우쳤으며 강민숙도 충분히 멀리 있는 상단 경로에 놓이지 않아 핵심 와이드 구도가 약하다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_꿈속_오솔길_464fe0.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:813104>"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S35sh1__bgfirst_bg.png",
   "bg_asset_id": "e18c9255-c6ce-4aef-8ed9-2f6097998213",
   "bg_record_key": "S35sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "꿈속 오솔길",
   "groupbg_asset_id": "575a0972-407e-47e4-ad05-b932d41fafa4"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S35sh7": {
  "input_fingerprint": "da1640e336e960c0",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 민숙의 몸이 완전히 검은 연기 형체로 흩어진 채 어둠 속에 고정된 순간.\n\nLOCATION (lock): The darkened upper end of the dream forest path where the mother's form dissolves into smoke. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: The dolly-out settles at chest height into a fixed wide hold, revealing the full dispersed outline of 강민숙(홍은미) (검은 연기 환영 상태) at the center of the upper path. Her human silhouette has broken into a smoke-like form with gaps showing the path behind it, while surrounding space remains visible so the transformation reads as a figure dissolving rather than a full-frame effect.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: upper forest path (daytime path within the dream) — The path recedes behind the central dissolving form; used as Shows through openings in the dispersing figure and establishes her former full-body position; forest (surrounding the path); used as Frames the transformation without obscuring the silhouette.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Daylight within the dream is held in ominous contrast against the black smoke-like transformation without introducing unsupported darkness or a visible source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Within the dream, Minsuk’s body remains dispersed as black smoke and her eyes retain their golden goat-like form.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강민숙(홍은미) (검은 연기 환영 상태) (Korean woman in her late 40s, dark hair, refined narrow face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 5,
   "C": 9
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "넓은 전신 연기 윤곽과 중앙 숲길은 잘 맞지만, 잠금된 황금 염소형 눈과 강민숙의 얼굴 정체성이 사라진 점이 결정적이다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "황금빛 눈은 보이지만 반신에 가까운 좁은 프레이밍과 온전한 얼굴·의상 때문에 ‘완전히 흩어진 전신 연기 형체를 보여주는 와이드 숏’을 충족하지 못한다."
   },
   {
    "label": "C",
    "score": 9,
    "verdict_ko": "중앙 전신 외곽, 연기 틈으로 보이는 후퇴하는 길, 넓은 숲 프레이밍, 황금 염소형 눈을 모두 가장 충실하게 구현했다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S35sh6_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "[치명적] 검은 연기 실루엣이 물리적 부피감 없이 평면적인 2D 그래픽 오버레이나 스티커처럼 렌더링되었으며, 몸통 부분에 정체불명의 납작한 황갈색 형태 두 개가 발생하여 '질감 현실성' 및 '임의의 사물 추가 금지' 규칙을 위반합니다.",
     "fix_en": "Replace the flat, 2D graphic-style black silhouette and the two solid tan artifacts in its torso with realistic, volumetric 3D black smoke that forms a human silhouette with natural gaps showing the background. Preserve the face, the forest path, the surrounding trees, the daylight lighting, and the exact wide-shot framing.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "[주요] 연기 속 얼굴이 기준 이미지의 40대 후반 여성에 비해 너무 젊고 매끄럽게 묘사되었으며, 요구된 염소 형태의 직사각형 눈동자 디테일이 보이지 않습니다.",
     "fix_en": "",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 2,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the flat, 2D graphic-style black silhouette and the two solid tan artifacts in its torso with realistic, volumetric 3D black smoke that forms a human silhouette with natural gaps showing the background. Preserve the face, the forest path, the surrounding trees, the daylight lighting, and the exact wide-shot framing.\n- \n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·장소·금빛 눈은 맞지만 몸통 중앙이 뻥 뚫리고 하반신이 잉크 튄 평면 얼룩으로 뭉개져 '인간 실루엣이 흩어진다'는 요구와 재질 사실성이 약하다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 고정 와이드·동일 장소·동일 금빛 염소 눈을 유지하면서 머리-어깨-팔-몸통의 인간 실루엣과 틈 사이로 보이는 길을 모두 살렸고, 연기가 평면 스플래터 대신 부피감 있는 물질로 읽혀 계약에 더 충실하다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.625,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.625,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·장소·금빛 눈은 맞지만 몸통 중앙이 뻥 뚫리고 하반신이 잉크 튄 평면 얼룩으로 뭉개져 '인간 실루엣이 흩어진다'는 요구와 재질 사실성이 약하다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 고정 와이드·동일 장소·동일 금빛 염소 눈을 유지하면서 머리-어깨-팔-몸통의 인간 실루엣과 틈 사이로 보이는 길을 모두 살렸고, 연기가 평면 스플래터 대신 부피감 있는 물질로 읽혀 계약에 더 충실하다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "가슴 높이 고정 와이드, 중앙의 흩어진 전신 실루엣, 틈 사이로 보이는 상단 숲길, 금빛 염소 눈 모두 충족하며 인체 형태를 최대한 유지해 '사람을 형체·덩어리로 축소하지 말라'는 조항에도 부합 — 표면 질감이 다소 굳은 점만 감점."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "연기를 더 투명하게 다듬어 '틈으로 길이 보인다'는 점은 살렸으나, 계약이 요구하지 않은 인체 볼륨 소실(어깨·팔·다리가 잉크 얼룩으로 붕괴)과 중경 대비 저하를 대가로 치러 '인간 형태를 최대한 렌더링' 조항에서 손실이 더 크다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.889,
      "B": 1.75
     },
     "adjusted": {
      "A": 1.889,
      "B": 1.75
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "B"
     },
     "agreed": false
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "가슴 높이 고정 와이드, 중앙의 흩어진 전신 실루엣, 틈 사이로 보이는 상단 숲길, 금빛 염소 눈 모두 충족하며 인체 형태를 최대한 유지해 '사람을 형체·덩어리로 축소하지 말라'는 조항에도 부합 — 표면 질감이 다소 굳은 점만 감점."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "연기를 더 투명하게 다듬어 '틈으로 길이 보인다'는 점은 살렸으나, 계약이 요구하지 않은 인체 볼륨 소실(어깨·팔·다리가 잉크 얼룩으로 붕괴)과 중경 대비 저하를 대가로 치러 '인간 형태를 최대한 렌더링' 조항에서 손실이 더 크다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev만 (배경 전용·공유 계획)",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S35sh6"
  },
  "lane_policy": "share_plan_prev_bgonly"
 },
 "S37sh3": {
  "input_fingerprint": "d2beed575b1dd1c4",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 열린 문 안쪽으로 낚시 용품과 식당 물품만이 가득 쌓인 평범한 창고 내부 전경.\n\nLOCATION (lock): Inside the ordinary backyard storage shed, now filled with restaurant supplies and fishing goods visible in daylight through the open door. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At eye level on the doorway centerline, the camera holds a static wide view from immediately outside the open storage entrance. Ordinary fishing equipment and restaurant supplies form readable layered stacks across the interior, with no person or altar visible.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: ordinary storage interior in the middle-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: 열린 창고 문과 출입구 (활짝 열림) — The opening faces the camera squarely, exposing the storage interior; used as Frames the full interior and confirms that the room is an ordinary storage space; 낚시 용품 (잔뜩 쌓여 있음); used as Supplies the principal visual evidence replacing the vanished altar; 식당 물품 (잔뜩 쌓여 있음); used as Fills the remaining storage depth with mundane restaurant inventory.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Daytime ambient light appropriate to the backyard and storage entrance keeps the contents plainly legible without introducing a visible lighting fixture.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The shed has been cleared of every visible trace of the night altar and now contains only restaurant stock and fishing supplies. Jihu’s necklace remains missing.\n\nPEOPLE: the SHOT TEXT alone decides who is visible in this shot. People known to appear somewhere in this scene: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face); 강영란(흰 머리 여자) (Korean woman in her late 60s, neatly twisted white hair, graceful dignified face). That list is scene-level, not a cast list for this frame — it may name someone this shot does not show, and it may omit someone this shot does show. If the shot text names a person who is not on the list, draw that person exactly as the shot text describes them; the list does not override the shot text. Never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 6,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "출입구 중심선·눈높이에서 문 바로 바깥에서 잡은 정면 와이드로 내부 전경을 정확히 담았고, 로케이션 사진의 블록 벽체·갈색 합판 문·마당 배치가 그대로 유지되며 낚시 용품과 식당 물품이 층층이 읽힌다. 소품 라벨 문자가 다소 뚜렷한 점만 감점.  ★위반: [gpt] 문짝의 안내 스티커와 상자에 읽히는 한글 인쇄가 있어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "내부 적재물의 층위 가독성과 주간 광선은 좋으나, 로케이션의 여닫이 문짝과 좁은 문틀이 사라지고 차고형 대형 개구부로 바뀌어 '활짝 열린 창고 문' 배경 요소와 구조 잠금이 약해졌다.  ★위반: [gpt] 상자와 음료 포장에 인쇄된 글자가 보여 ‘어디에도 텍스트 금지’ 조건을 위반한다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "로케이션 사진의 카메라 프레이밍을 사실상 복제해 '문 바로 바깥에서의 내부 전경 와이드'가 아닌 마당 원경이 되었고, 블록 벽체·문짝이 사라진 얕은 판넬 상자형 구조로 구조 잠금도 크게 어긋난다.  ★위반: [gpt] 포대, 상자와 전면 용기에 읽히는 인쇄 문구가 있어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L22B03.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_wooden_restaurant_compound_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "창고 건물이 지정된 STRUCTURE LOOK 레퍼런스(흰색 나무 외벽, 구조, 비율)를 따르지 않고 LOCATION 레퍼런스의 콘크리트 블록 건물로 잘못 렌더링되었습니다.",
     "fix_en": "Change the concrete block walls of the shed to vertical white wooden siding matching the reference, preserving the camera angle, daylight lighting, the open door, the interior contents, and the surrounding garden.",
     "needs_regeneration": true
    },
    {
     "issue_ko": "창고 내부 맨 위 선반에 있는 종이 상자 표면에 텍스트가 렌더링되었습니다 (프롬프트의 텍스트 금지 지시 위반).",
     "fix_en": "Remove all text and logos from the cardboard boxes on the top shelf, making them plain brown cardboard, while preserving the boxes' shapes, all other supplies, the shed interior, and the current framing."
    },
    {
     "issue_ko": "열린 문 안쪽 표면에 텍스트가 적힌 안내문/스티커가 있습니다 (텍스트 금지 지시 위반).",
     "fix_en": "Remove the green paper and text from the inside surface of the open door, replacing it with bare wood grain, while preserving the door itself, the fishing net hanging on it, the shed interior, and the framing."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the concrete block walls of the shed to vertical white wooden siding matching the reference, preserving the camera angle, daylight lighting, the open door, the interior contents, and the surrounding garden.\n- Remove all text and logos from the cardboard boxes on the top shelf, making them plain brown cardboard, while preserving the boxes' shapes, all other supplies, the shed interior, and the current framing.\n- Remove the green paper and text from the inside surface of the open door, replacing it with bare wood grain, while preserving the door itself, the fishing net hanging on it, the shed interior, and the framing.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·적재물·일광은 계약대로지만, 상단 박스의 한글 인쇄 라벨과 문짝 스티커가 '텍스트·캡션·주석 전면 금지'를 정면으로 위반한다.  ★위반: [claude-opus] 프레임 내 판독 가능한 인쇄 문자(상단 박스의 '재료제료' 등 한글 라벨, 문짝의 초록 스티커) — '어떤 텍스트·캡션·주석도 금지' 위반 / [gpt] 상자와 문짝 스티커에 선명한 한글 문구가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 정면 와이드 구도와 낚시·식당 물품 적층을 유지하면서 금지된 인쇄 문자를 무지 박스로 제거했고, 외벽만 STRUCTURE LOOK의 판벽으로 바뀐 것이 유일한 비용이라 이득이 손실을 앞선다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.25,
      "B": 2.0
     },
     "adjusted": {
      "A": 0.75,
      "B": 2.0
     },
     "violations": {
      "A": [
       "[claude-opus] 프레임 내 판독 가능한 인쇄 문자(상단 박스의 '재료제료' 등 한글 라벨, 문짝의 초록 스티커) — '어떤 텍스트·캡션·주석도 금지' 위반",
       "[gpt] 상자와 문짝 스티커에 선명한 한글 문구가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·적재물·일광은 계약대로지만, 상단 박스의 한글 인쇄 라벨과 문짝 스티커가 '텍스트·캡션·주석 전면 금지'를 정면으로 위반한다.  ★위반: [claude-opus] 프레임 내 판독 가능한 인쇄 문자(상단 박스의 '재료제료' 등 한글 라벨, 문짝의 초록 스티커) — '어떤 텍스트·캡션·주석도 금지' 위반 / [gpt] 상자와 문짝 스티커에 선명한 한글 문구가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 정면 와이드 구도와 낚시·식당 물품 적층을 유지하면서 금지된 인쇄 문자를 무지 박스로 제거했고, 외벽만 STRUCTURE LOOK의 판벽으로 바뀐 것이 유일한 비용이라 이득이 손실을 앞선다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "all_candidates_fail": false,
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 열린 문 정면, 출입구 중심선상 눈높이에서 창고 내부를 정면으로 바라본다. 열린 나무문은 왼쪽으로 활짝 젖혀져 있고 시선은 내부 선반 더미로 곧게 들어간다. 인물·시선·총구 등 겨냥 대상은 없으며(사람 없음), 낚싯대 끝은 위쪽 문틀 방향으로 기대어 있어 지지면과 일치한다.",
      "built_space": "창고는 개구부 1개, 여닫이 문 1짝(경첩 왼쪽), 지붕은 슬레이트 처마가 프레임 상단에 걸림. 다만 외벽이 흰색 판재(나무)로 렌더링되어 LOCATION 사진의 시멘트 블록 벽과 재질이 어긋난다. 내부는 좌측 낚시용품 선반, 중앙·우측 식당 식기/그릇 선반으로 층위가 명확. 선반 개수·배치는 자연스럽고 중복 구조물 없음.",
      "entities": "낚시용품: 뜰채, 낚싯대 다발, 릴, 태클박스, 웨이더 — 모두 실제 낚시 장비로 읽힘. 식당 물품: 접시 더미, 냄비, 대형 플라스틱 통, 양념통, 종이박스 — 식당 재고로 읽힘. 사람 없음(계약대로), 제단·목걸이 흔적 없음. 문자·라벨은 거의 판독 불가 수준의 미세한 것뿐이라 '텍스트 금지' 조항에 안전.",
      "hard_violations": [],
      "physics": "모든 물체가 선반, 바닥, 문짝, 벽에 접지되어 있다. 뜰채는 문 안쪽에 걸려 손잡이가 바닥에 닿아 기댐, 낚싯대는 문틀/벽에 기대어 하단이 바닥에 접함, 접시 더미는 선반 위에 안착. 공중 부양 물체 없음. 그림자 방향도 주광 일광과 일치."
     },
     {
      "label": "B",
      "direction": "동일하게 출입구 정면 눈높이 정적 와이드. 시선 축은 내부 중앙 선반으로 곧게 향함. 인물 없음. 낚싯대·뜰채 방향은 A와 동일하게 기댐 방향과 일치.",
      "built_space": "창고 외벽이 시멘트 블록으로 렌더링되어 LOCATION 사진과 재질이 더 잘 맞는다. 개구부 1개, 문 1짝, 처마 1개로 구조 중복 없음. 내부 레이아웃은 A와 사실상 동일.",
      "entities": "낚시용품·식당 물품 구성은 A와 동일. 그러나 상단 종이박스 두 개에 한글 인쇄 라벨('재료제료' 등 왜곡된 글자)이 크게 찍혀 있고, 문 안쪽에 초록색 안내판/스티커가 추가되어 있으며 그 글자들이 뭉개진 가짜 텍스트다. 계약의 '어떤 텍스트·캡션·주석도 금지' 조항에 정면 위배되며, 스티커는 표면에 붙은 그래픽 오버레이처럼 읽힌다.",
      "hard_violations": [
       "프레임 내 판독 가능한 인쇄 텍스트(박스 라벨 '재료제료', 문 안쪽 초록 안내판 글자) — '텍스트·캡션·주석 금지' 조항 위반 및 왜곡된 문자로 유출된 마커성 요소"
      ],
      "physics": "모든 물체가 선반·바닥·문에 접지. 부양 물체 없음. 다만 문 안쪽 초록 표지판은 두께·광택 없이 평면 스티커처럼 얹혀 있어 재질 사실성이 약함."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "출입구 정면 눈높이 정적 와이드, 사람·제단 없이 낚시용품과 식당 물품이 층층이 읽히는 구성을 정확히 충족하고 프레임 내 판독 가능한 문자가 없다; 외벽이 블록이 아닌 판재로 보이는 점은 감점이지만 계약의 명시 금지 조항을 어기지 않아 우위."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "외벽을 시멘트 블록으로 되돌린 이득은 있으나, 계약이 요구하지 않은 한글 인쇄 박스 라벨과 문 안쪽 초록 안내판을 새로 넣어 '텍스트·캡션·주석 금지'를 정면으로 위반했고 글자도 뭉개진 가짜라 손실이 이득을 넘는다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "출입구 정면 눈높이 정적 와이드, 사람·제단 없이 낚시용품과 식당 물품이 층층이 읽히는 구성을 정확히 충족하고 프레임 내 판독 가능한 문자가 없다; 외벽이 블록이 아닌 판재로 보이는 점은 감점이지만 계약의 명시 금지 조항을 어기지 않아 우위."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "외벽을 시멘트 블록으로 되돌린 이득은 있으나, 계약이 요구하지 않은 한글 인쇄 박스 라벨과 문 안쪽 초록 안내판을 새로 넣어 '텍스트·캡션·주석 금지'를 정면으로 위반했고 글자도 뭉개진 가짜라 손실이 이득을 넘는다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "카메라는 열린 문 정면, 출입구 중심선상 눈높이에서 창고 내부를 정면으로 바라본다. 열린 나무문은 왼쪽으로 활짝 젖혀져 있고 시선은 내부 선반 더미로 곧게 들어간다. 인물·시선·총구 등 겨냥 대상은 없으며(사람 없음), 낚싯대 끝은 위쪽 문틀 방향으로 기대어 있어 지지면과 일치한다.",
      "built_space": "창고는 개구부 1개, 여닫이 문 1짝(경첩 왼쪽), 지붕은 슬레이트 처마가 프레임 상단에 걸림. 다만 외벽이 흰색 판재(나무)로 렌더링되어 LOCATION 사진의 시멘트 블록 벽과 재질이 어긋난다. 내부는 좌측 낚시용품 선반, 중앙·우측 식당 식기/그릇 선반으로 층위가 명확. 선반 개수·배치는 자연스럽고 중복 구조물 없음.",
      "entities": "낚시용품: 뜰채, 낚싯대 다발, 릴, 태클박스, 웨이더 — 모두 실제 낚시 장비로 읽힘. 식당 물품: 접시 더미, 냄비, 대형 플라스틱 통, 양념통, 종이박스 — 식당 재고로 읽힘. 사람 없음(계약대로), 제단·목걸이 흔적 없음. 문자·라벨은 거의 판독 불가 수준의 미세한 것뿐이라 '텍스트 금지' 조항에 안전.",
      "hard_violations": [],
      "physics": "모든 물체가 선반, 바닥, 문짝, 벽에 접지되어 있다. 뜰채는 문 안쪽에 걸려 손잡이가 바닥에 닿아 기댐, 낚싯대는 문틀/벽에 기대어 하단이 바닥에 접함, 접시 더미는 선반 위에 안착. 공중 부양 물체 없음. 그림자 방향도 주광 일광과 일치."
     },
     {
      "label": "A",
      "direction": "동일하게 출입구 정면 눈높이 정적 와이드. 시선 축은 내부 중앙 선반으로 곧게 향함. 인물 없음. 낚싯대·뜰채 방향은 A와 동일하게 기댐 방향과 일치.",
      "built_space": "창고 외벽이 시멘트 블록으로 렌더링되어 LOCATION 사진과 재질이 더 잘 맞는다. 개구부 1개, 문 1짝, 처마 1개로 구조 중복 없음. 내부 레이아웃은 A와 사실상 동일.",
      "entities": "낚시용품·식당 물품 구성은 A와 동일. 그러나 상단 종이박스 두 개에 한글 인쇄 라벨('재료제료' 등 왜곡된 글자)이 크게 찍혀 있고, 문 안쪽에 초록색 안내판/스티커가 추가되어 있으며 그 글자들이 뭉개진 가짜 텍스트다. 계약의 '어떤 텍스트·캡션·주석도 금지' 조항에 정면 위배되며, 스티커는 표면에 붙은 그래픽 오버레이처럼 읽힌다.",
      "hard_violations": [
       "프레임 내 판독 가능한 인쇄 텍스트(박스 라벨 '재료제료', 문 안쪽 초록 안내판 글자) — '텍스트·캡션·주석 금지' 조항 위반 및 왜곡된 문자로 유출된 마커성 요소"
      ],
      "physics": "모든 물체가 선반·바닥·문에 접지. 부양 물체 없음. 다만 문 안쪽 초록 표지판은 두께·광택 없이 평면 스티커처럼 얹혀 있어 재질 사실성이 약함."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "플레이트+seed만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_bypass:bg_only"
 },
 "S39sh10": {
  "input_fingerprint": "964bde580f32cfa3",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 엄격한 표정으로 숲 출입 금지를 경고하는 금월도 마을 주민들 앞에서 차마 입을 떼지 못하고 얼어붙은 김지후(강수리영)의 측면.\n\nLOCATION (lock): A house-front section of the coastal village lane near the route toward the forest. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height beside Suri-young, the static camera holds her in a strict profile at medium distance, with the warning residents positioned across from her in the same frame. She is caught after attempting to leave—weight arrested on the leading foot and lips sealed—while the residents' rigid faces establish the source of the warning.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 야트막한 돌담 (주민들이 담 너머에 있음) — The wall extends behind and beneath the residents across from Suri-young; used as Separates the warning residents from Suri-young while preserving their face-to-face spatial relationship; 마을 길 (낮의 마을 길) — The path direction continues past Suri-young's arrested leading foot; used as Frames the place from which Suri-young was trying to depart.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime ambient light is rendered with firm, restrained contrast that hardens the residents' warning and Suri-young's frozen profile.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same coastal lane, low walls, clustered residents, daylight, and persistent plant-filled yards. Exclude the zoo-like crowd emphasis; frame the residents warning the woman against entering the forest.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains without her animal-bone necklace as the villagers warn her.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "수리영은 화면 오른쪽에서 왼쪽을 향한 좌향 측면으로 서 있고 시선은 담 너머 주민들 쪽을 향한다. 주민들(6명)은 담 너머에서 대체로 카메라/왼쪽 전방을 보는데, 여러 명의 시선이 수리영이 아니라 렌즈 방향(정면)으로 향해 있어 경고의 대상 지향이 약하다. 전체 구도는 서로 마주 본 관계가 성립하긴 하나 두 그룹 사이 거리가 매우 멀다.",
    "built_space": "이전 스틸의 집 앞 구간, 낮은 현무암 돌담, 나무 대문, 걸린 마른 식물, 사다리, 전신주, 바다 방향 원경이 거의 그대로 유지된다. 주민들은 담 너머 마당 쪽에 위치해 담이 두 진영을 분리하는 조건은 충족. 다만 프레임이 이전 스틸의 와이드 앵글을 거의 그대로 복제해 배경 비중이 지나치게 크다.",
    "entities": "수리영: 20대 초반 한국 여성, 긴 검은 머리, 회색 캡, 갈색 워크재킷, 짙은 카고 바지, 갈색 부츠 — 캐릭터 레퍼런스와 일치. 뼈 목걸이 없음(계약 준수). 주민 6명은 이전 스틸의 동일 인물·복장. 텍스트/워터마크 없음.",
    "physics": "수리영은 두 발 모두 지면 접지, 앞발에 체중이 실린 걸음 중지 자세로 읽힌다. 주민들은 담에 팔을 얹거나 담 뒤에 서 있어 지지 명확. 공중 부양 요소 없음.",
    "hard_violations": []
   },
   {
    "label": "B",
    "direction": "수리영은 화면 우측에 좌향 엄격한 측면(프로필)으로 서서 시선이 담 너머 주민들에게 정확히 꽂힌다. 주민들 6명 모두 얼굴이 오른쪽(수리영 쪽)으로 돌아가 있고 시선이 그녀를 향한다 — 경고의 방향이 서로 맞물린다.",
    "built_space": "낮은 현무암 돌담이 주민들 뒤·아래로 이어지고 주민들은 담 너머에 서 있어 분리+대면 관계가 명확. 나무 문기둥, 처마, 덩굴 식물, 기와지붕, 전신주, 바다 방향 원경 등 이전 스틸 요소가 유지된다. 마을 길은 수리영의 멈춘 발 옆으로 계속 이어져 떠나려던 방향을 보여준다.",
    "entities": "수리영: 20대 초반 한국 여성, 긴 검은 생머리, 갈색 워크재킷, 회색 카고 바지 — 레퍼런스 얼굴/체형과 부합. 모자는 벗은 상태로 레퍼런스의 캡이 없는 점이 소소한 차이. 뼈 목걸이 없음(준수). 주민은 이전 스틸의 6인과 동일 인상. 문자/워터마크 없음.",
    "physics": "수리영 두 발 지면 접지, 앞발에 체중이 실린 정지 상태, 팔은 자연스럽게 내려짐. 주민들은 담에 손을 얹거나 담 뒤 지면에 서 있음. 부양·무지지 요소 없음.",
    "hard_violations": []
   },
   {
    "label": "C",
    "direction": "수리영은 우측에서 좌향 측면으로 서서 시선은 앞(왼쪽)을 향하나 주민들보다 살짝 아래/전방으로 빠져 정확한 눈맞춤은 약간 어긋난다. 주민 5~6명은 담 너머에서 그녀 쪽(오른쪽 전방)을 응시 — 대체로 방향은 맞다.",
    "built_space": "돌담, 나무 기둥, 사다리, 마른 식물, 집 처마 등 로케이션 요소 유지. 다만 주민들이 담 너머 마당의 낮은 위치가 아니라 담 옆/뒤에 서 있어 담이 분리선 역할을 하는 정도가 약하고, 원경 항구와 배가 다소 확대되어 보인다. 프레임은 미디엄보다 넓은 편.",
    "entities": "수리영: 20대 초반 한국 여성, 긴 검은 머리, 갈색 재킷, 짙은 카고 바지 — 레퍼런스와 부합(모자 없음). 뼈 목걸이 없음. 주민 인물은 이전 스틸 인물과 유사하나 복장 색·배치가 일부 달라짐.",
    "physics": "수리영 양발 지면 접지, 다만 발이 거의 모아진 차렷에 가까워 '떠나려다 멈춘 앞발 체중' 표현이 약하다. 주민들 모두 지면 접지. 부양 요소 없음.",
    "hard_violations": []
   }
  ],
  "totals": {
   "A": 5,
   "B": 9,
   "C": 6
  },
  "selected": "B",
  "ranking": [
   "B",
   "C",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "이전 스틸의 와이드 프레임을 그대로 답습해 '미디엄 숏·가슴 높이 측면'이라는 카메라 지시를 어겼고, 주민 다수가 수리영이 아닌 렌즈를 보고 있어 경고의 시선 방향도 흐릿하다."
   },
   {
    "label": "B",
    "score": 9,
    "verdict_ko": "가슴 높이 정측면 미디엄 숏, 담 너머 주민 전원이 수리영을 노려보는 대면 관계, 멈춘 앞발과 다문 입술까지 샷 텍스트를 가장 정확히 구현했다."
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "로케이션과 인물은 유지되나 프레임이 다소 넓고 주민이 담 너머가 아닌 옆에 서 있어 분리 기능이 약하며, 수리영의 자세가 차렷에 가까워 '떠나려다 얼어붙은' 순간이 살지 않는다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S39sh7_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:813104>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 샷과 비교해 주민들이 서 있는 위치와 순서가 완전히 달라졌으며(할머니와 초록 비니 남성의 위치가 바뀜), 대문과 식물 건조대 등 담장 주변의 기존 구조물들이 누락되었습니다.",
     "fix_en": "Restore the missing wooden gate and the hanging green plants to the stone wall in front of the leftmost villagers, preserving the current characters, their poses, the lighting, and the framing exactly as they are.",
     "needs_regeneration": true
    },
    {
     "issue_ko": "우측의 김지후가 캐릭터 레퍼런스에서 착용하고 있던 회색 캡 모자가 누락되었습니다.",
     "fix_en": "Add the dark grey cap from the character reference onto the head of the woman on the right, keeping her face, long loose black hair, clothing, the villagers, and the background exactly unchanged."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore the missing wooden gate and the hanging green plants to the stone wall in front of the leftmost villagers, preserving the current characters, their poses, the lighting, and the framing exactly as they are.\n- Add the dark grey cap from the character reference onto the head of the woman on the right, keeping her face, long loose black hair, clothing, the villagers, and the background exactly unchanged.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 화면 오른쪽에서 좌측을 향한 엄격한 측면(프로필)으로 서 있고 시선은 담 너머 주민들 쪽으로 향한다. 주민 5명(할머니, 비니 청년, 중년 남성, 여성, 안경 없는 중년 남성)은 모두 오른쪽 지후 방향으로 시선을 보내며 경고하는 표정. 시선 축이 서로 맞물려 대치가 성립한다.",
      "built_space": "왼쪽에 현무암 돌담이 길게 이어지고 주민들은 담 너머 마당 쪽에 있음 — 참조 스틸의 배치와 일치. 다만 참조 스틸에 있던 담 위/마당의 무성한 초록 식물(널려 자란 덩굴·잎)이 대부분 사라지고 맨 돌담이 노출됐다. 나무 사다리(검은 프레임)도 사라짐. 마을 길은 지후의 앞발 방향으로 계속 이어져 프레임 우측 심도를 만든다.",
      "entities": "김지후: 20대 초반 한국 여성, 긴 검은 생머리, 갈색 워크재킷+짙은 카고팬츠로 캐릭터 레퍼런스와 일치. 그러나 레퍼런스의 회색 볼캡을 쓰고 있지 않다(모자 없음). 동물뼈 목걸이 없음 — CARRIED STATE 충족. 주민들 의상은 이전 스틸과 대체로 일치(할머니 보라 저고리, 비니 청년, 중년 남녀).",
      "physics": "지후는 두 발 모두 지면에 접지, 선행 발에 체중이 실린 정지 자세로 읽힘. 주민들은 돌담에 손을 얹거나 담 뒤에 서 있어 지지가 명확. 공중에 뜬 것 없음.",
      "hard_violations": []
     },
     {
      "label": "B",
      "direction": "김지후는 우측에서 좌향 측면으로 서고 시선은 주민들 쪽으로 고정. 주민 5명 모두 지후 쪽을 응시하며 경고하는 굳은 표정. 시선 대치 성립, 샷 텍스트와 일치.",
      "built_space": "왼쪽 돌담과 나무 대문, 담 위로 널린 초록 잎사귀 다발, 검은 금속 사다리까지 이전 스틸의 고정 요소가 그대로 유지됨. 주민들은 담 너머에 위치. 마을 길은 지후 앞쪽으로 이어져 떠나려던 방향을 프레임한다. 담 높이·재질·마당 식생 모두 레퍼런스와 부합.",
      "entities": "김지후: 20대 초반 한국 여성, 긴 검은 생머리, 회색 볼캡, 갈색 워크재킷, 짙은 카고팬츠 — 캐릭터 레퍼런스와 정확히 일치. 뼈 목걸이 없음 — CARRIED STATE 충족. 주민 5인 의상·얼굴 이전 스틸과 일치.",
      "physics": "지후는 양발 접지, 앞발에 체중이 멈춘 정지 포즈. 팔은 자연스럽게 내려 몸통에 붙음. 주민들은 담·대문에 기대거나 손을 얹고 있어 지지 명확. 부유물 없음.",
      "hard_violations": []
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "측면 대치 구도와 시선은 맞지만, 계약이 요구하지 않은 변경이 크다 — 캐릭터 레퍼런스의 회색 볼캡이 사라졌고, 이전 스틸에 잠긴 담 위 초록 식물 다발과 사다리가 제거되어 '식물 가득한 마당 유지' 지시를 어겼다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "체스트 하이 정측면 미디엄샷, 담 너머 주민들과의 대치, 뼈 목걸이 부재까지 충족하면서 레퍼런스의 볼캡·담 위 식생·사다리 등 고정 요소를 그대로 보존해 변경이 가장 적다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "측면 대치 구도와 시선은 맞지만, 계약이 요구하지 않은 변경이 크다 — 캐릭터 레퍼런스의 회색 볼캡이 사라졌고, 이전 스틸에 잠긴 담 위 초록 식물 다발과 사다리가 제거되어 '식물 가득한 마당 유지' 지시를 어겼다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "체스트 하이 정측면 미디엄샷, 담 너머 주민들과의 대치, 뼈 목걸이 부재까지 충족하면서 레퍼런스의 볼캡·담 위 식생·사다리 등 고정 요소를 그대로 보존해 변경이 가장 적다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 화면 오른쪽에서 좌측을 향한 엄격한 측면(프로필)으로 서 있고 시선은 담 너머 주민들 쪽으로 향한다. 주민 5명(할머니, 비니 청년, 중년 남성, 여성, 안경 없는 중년 남성)은 모두 오른쪽 지후 방향으로 시선을 보내며 경고하는 표정. 시선 축이 서로 맞물려 대치가 성립한다.",
      "built_space": "왼쪽에 현무암 돌담이 길게 이어지고 주민들은 담 너머 마당 쪽에 있음 — 참조 스틸의 배치와 일치. 다만 참조 스틸에 있던 담 위/마당의 무성한 초록 식물(널려 자란 덩굴·잎)이 대부분 사라지고 맨 돌담이 노출됐다. 나무 사다리(검은 프레임)도 사라짐. 마을 길은 지후의 앞발 방향으로 계속 이어져 프레임 우측 심도를 만든다.",
      "entities": "김지후: 20대 초반 한국 여성, 긴 검은 생머리, 갈색 워크재킷+짙은 카고팬츠로 캐릭터 레퍼런스와 일치. 그러나 레퍼런스의 회색 볼캡을 쓰고 있지 않다(모자 없음). 동물뼈 목걸이 없음 — CARRIED STATE 충족. 주민들 의상은 이전 스틸과 대체로 일치(할머니 보라 저고리, 비니 청년, 중년 남녀).",
      "physics": "지후는 두 발 모두 지면에 접지, 선행 발에 체중이 실린 정지 자세로 읽힘. 주민들은 돌담에 손을 얹거나 담 뒤에 서 있어 지지가 명확. 공중에 뜬 것 없음.",
      "hard_violations": []
     },
     {
      "label": "B",
      "direction": "김지후는 우측에서 좌향 측면으로 서고 시선은 주민들 쪽으로 고정. 주민 5명 모두 지후 쪽을 응시하며 경고하는 굳은 표정. 시선 대치 성립, 샷 텍스트와 일치.",
      "built_space": "왼쪽 돌담과 나무 대문, 담 위로 널린 초록 잎사귀 다발, 검은 금속 사다리까지 이전 스틸의 고정 요소가 그대로 유지됨. 주민들은 담 너머에 위치. 마을 길은 지후 앞쪽으로 이어져 떠나려던 방향을 프레임한다. 담 높이·재질·마당 식생 모두 레퍼런스와 부합.",
      "entities": "김지후: 20대 초반 한국 여성, 긴 검은 생머리, 회색 볼캡, 갈색 워크재킷, 짙은 카고팬츠 — 캐릭터 레퍼런스와 정확히 일치. 뼈 목걸이 없음 — CARRIED STATE 충족. 주민 5인 의상·얼굴 이전 스틸과 일치.",
      "physics": "지후는 양발 접지, 앞발에 체중이 멈춘 정지 포즈. 팔은 자연스럽게 내려 몸통에 붙음. 주민들은 담·대문에 기대거나 손을 얹고 있어 지지 명확. 부유물 없음.",
      "hard_violations": []
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "지후는 화면 오른쪽에서 왼쪽을 향한 엄격한 측면(프로필)으로 서 있고, 시선은 돌담 너머 주민들 쪽으로 정확히 향한다. 주민 5인(할머니, 비니 쓴 청년, 중년 남성, 중년 여성, 안경 없는 남성)은 모두 오른쪽 지후 방향을 응시하며 경고하는 표정. 시선 축이 서로 맞물려 대면 관계가 성립한다. 마을 길은 지후의 선행 발 방향(화면 오른쪽 안쪽)으로 계속 이어진다.",
      "built_space": "레퍼런스의 현무암 돌담, 나무 대문(세로 판자문), 담에 기대 세운 검은 사다리/사각 지지대, 담 위에 널린 초록 식물 다발, 기와지붕 집, 전신주와 전선, 원경 바다와 부두까지 이전 스틸의 고정 요소가 대부분 유지됨. 담은 주민들 앞·아래로 이어져 지후와 분리하면서 대면 배치를 유지. 벽·문·사다리 개수 중복 없음.",
      "entities": "지후: 20대 초반 한국 여성, 긴 검은 머리, 회색 캡, 갈색 워크 재킷, 다크 카고 팬츠 — 캐릭터 레퍼런스와 일치. 목에는 뼈 목걸이 없음(CARRIED STATE 충족). 주민들은 이전 스틸과 동일 인물·동일 복장(할머니 보라 자주 상의, 비니 청년, 회색 셔츠 남성 등)으로 유지. 텍스트·워터마크 없음.",
      "physics": "지후는 양발이 지면에 있고 선행 발에 체중이 걸린 채 멈춘 자세, 그림자가 발 밑에 정확히 붙음. 주민들은 담에 손·팔을 얹거나 담 뒤 지면에 서 있어 지지가 명확. 공중에 떠 있는 인물이나 물체 없음.",
      "hard_violations": []
     },
     {
      "label": "B",
      "direction": "지후는 오른쪽에서 왼쪽을 보는 측면으로 주민들을 응시하고, 주민 5인도 지후 쪽을 향해 경고하는 시선. 대면 축은 성립. 길 방향도 유지.",
      "built_space": "돌담이 프레임 전체를 가로질러 크게 확장되고 높아졌으며, 이전 스틸에 있던 담 위 초록 식물 다발·담쟁이 덤불·검은 사다리(사각 지지대)가 사라졌다. 나무 대문은 판자문에서 좁은 기둥형으로 축소·변형. 결과적으로 '식물 가득한 마당(persistent plant-filled yards)'이라는 계승 지시가 상당 부분 소실되고, 돌담이 전경을 과하게 점유한다. 중복 구조물은 없음.",
      "entities": "지후: 캐릭터 레퍼런스와 얼굴·긴 검은 머리는 유사하나 회색 캡(모자)이 제거되어 맨머리로 렌더됨 — 이전 스틸/캐릭터 레퍼런스의 의상 상태 유지 위반. 재킷·카고 팬츠는 일치, 뼈 목걸이 없음(충족). 주민들은 인물·복장 유지되나 상반신 노출이 담에 더 가려짐. 텍스트 없음.",
      "physics": "지후는 두 발 지면 접지, 그림자 일치, 정지 자세로 지지 문제 없음. 주민들은 담에 손을 얹은 채 서 있어 지지 명확. 공중 부양 요소 없음.",
      "hard_violations": []
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "가슴 높이 정측면·미디엄 숏, 담 너머 주민 대면, 캡 포함 캐릭터 레퍼런스 의상과 이전 스틸의 식물·나무문·사다리 등 고정 요소를 모두 유지해 계승 지시를 가장 충실히 충족한다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "구도·시선은 맞지만 계약이 요구하지 않은 변경이 크다 — 지후의 회색 캡이 사라졌고 담 위 식물 다발·덤불·사다리가 삭제되며 돌담이 전경으로 확대되어 '식물 가득한 마당' 계승 조항과 인물 복장 일관성을 잃었다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "가슴 높이 정측면·미디엄 숏, 담 너머 주민 대면, 캡 포함 캐릭터 레퍼런스 의상과 이전 스틸의 식물·나무문·사다리 등 고정 요소를 모두 유지해 계승 지시를 가장 충실히 충족한다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "구도·시선은 맞지만 계약이 요구하지 않은 변경이 크다 — 지후의 회색 캡이 사라졌고 담 위 식물 다발·덤불·사다리가 삭제되며 돌담이 전경으로 확대되어 '식물 가득한 마당' 계승 조항과 인물 복장 일관성을 잃었다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "지후는 화면 오른쪽에서 왼쪽을 향한 엄격한 측면(프로필)으로 서 있고, 시선은 돌담 너머 주민들 쪽으로 정확히 향한다. 주민 5인(할머니, 비니 쓴 청년, 중년 남성, 중년 여성, 안경 없는 남성)은 모두 오른쪽 지후 방향을 응시하며 경고하는 표정. 시선 축이 서로 맞물려 대면 관계가 성립한다. 마을 길은 지후의 선행 발 방향(화면 오른쪽 안쪽)으로 계속 이어진다.",
      "built_space": "레퍼런스의 현무암 돌담, 나무 대문(세로 판자문), 담에 기대 세운 검은 사다리/사각 지지대, 담 위에 널린 초록 식물 다발, 기와지붕 집, 전신주와 전선, 원경 바다와 부두까지 이전 스틸의 고정 요소가 대부분 유지됨. 담은 주민들 앞·아래로 이어져 지후와 분리하면서 대면 배치를 유지. 벽·문·사다리 개수 중복 없음.",
      "entities": "지후: 20대 초반 한국 여성, 긴 검은 머리, 회색 캡, 갈색 워크 재킷, 다크 카고 팬츠 — 캐릭터 레퍼런스와 일치. 목에는 뼈 목걸이 없음(CARRIED STATE 충족). 주민들은 이전 스틸과 동일 인물·동일 복장(할머니 보라 자주 상의, 비니 청년, 회색 셔츠 남성 등)으로 유지. 텍스트·워터마크 없음.",
      "physics": "지후는 양발이 지면에 있고 선행 발에 체중이 걸린 채 멈춘 자세, 그림자가 발 밑에 정확히 붙음. 주민들은 담에 손·팔을 얹거나 담 뒤 지면에 서 있어 지지가 명확. 공중에 떠 있는 인물이나 물체 없음.",
      "hard_violations": []
     },
     {
      "label": "A",
      "direction": "지후는 오른쪽에서 왼쪽을 보는 측면으로 주민들을 응시하고, 주민 5인도 지후 쪽을 향해 경고하는 시선. 대면 축은 성립. 길 방향도 유지.",
      "built_space": "돌담이 프레임 전체를 가로질러 크게 확장되고 높아졌으며, 이전 스틸에 있던 담 위 초록 식물 다발·담쟁이 덤불·검은 사다리(사각 지지대)가 사라졌다. 나무 대문은 판자문에서 좁은 기둥형으로 축소·변형. 결과적으로 '식물 가득한 마당(persistent plant-filled yards)'이라는 계승 지시가 상당 부분 소실되고, 돌담이 전경을 과하게 점유한다. 중복 구조물은 없음.",
      "entities": "지후: 캐릭터 레퍼런스와 얼굴·긴 검은 머리는 유사하나 회색 캡(모자)이 제거되어 맨머리로 렌더됨 — 이전 스틸/캐릭터 레퍼런스의 의상 상태 유지 위반. 재킷·카고 팬츠는 일치, 뼈 목걸이 없음(충족). 주민들은 인물·복장 유지되나 상반신 노출이 담에 더 가려짐. 텍스트 없음.",
      "physics": "지후는 두 발 지면 접지, 그림자 일치, 정지 자세로 지지 문제 없음. 주민들은 담에 손을 얹은 채 서 있어 지지 명확. 공중 부양 요소 없음.",
      "hard_violations": []
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 17
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S39sh7"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S40sh5": {
  "input_fingerprint": "2e5270ccfce3c634",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 김여상의 몸이 비명을 지르듯 입을 크게 벌린 채 어두운 비탈길 아래로 기울어져 허공에 떠 있는 과거 기억 속 떨어지는 mid-action 찰나.\n\nLOCATION (lock): A steep, dark forest slope beside the island footpath, recalled from the previous night. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From a medium distance at mid-body height, the camera looks perpendicular across the slope and tilts with 김여상's downward trajectory, catching him in three-quarter view as his torso pitches away from the path. His suspended body occupies the middle-right of the frame while the path remains along the upper edge, preserving the instant before he drops from sight.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: 김여상 in the middle-right of the frame, midground, moves toward below the path; upper path edge in the upper-center of the frame, background.\n- KEY BACKGROUND ELEMENTS: forest path edge (Above 김여상's falling body); used as Upper-frame spatial reference showing where the fall began; slope below the path (김여상 is falling down it); used as Diagonal background establishing the direction of the fall.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key nighttime illumination preserves 김여상's silhouette and strained expression against the darker slope.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김여상 (Korean man in his mid-40s, dark hair, ordinary mature face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 6,
   "B": 1,
   "C": 8
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "의상·장소·낙하 순간의 물리적 지지는 무난하나, 인물이 화면 중앙~좌측에 놓이고 낙하 방향도 좌하향이라 '중앙 우측·비탈 아래로' 프레임 레이아웃 지시에서 벗어난다."
   },
   {
    "label": "B",
    "score": 1,
    "verdict_ko": "지면이 화면 상단 천장으로 뒤집힌 전도 구도에 인물은 아무 지지 없이 허공에 떠 있어 물리·카메라 지시 모두 치명적으로 위반; 미디엄 숏도 아닌 원경 크기다.  ★위반: [claude-opus] 인물이 어떤 지지도 없이 완전히 허공에 수평으로 떠 있고, 도약이나 착지 지점이 전혀 없음(부유체) / [claude-opus] 지면(숲길)이 화면 상단 천장으로 뒤집혀 렌더되어 중력·공간이 물리적으로 불가능한 전도 구도 / [gpt] 김여상의 몸이 길이나 비탈과 분리된 채 숲 공간에 수평으로 떠 있고, 추락의 출발 접점이나 도달할 경사면이 보이지 않아 물리적으로 지지되지 않는다."
   },
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "비탈을 가로지르는 미디엄 숏, 중앙 우측 3/4 각도, 상단의 길 가장자리와 대각선 사면, 입을 벌린 표정과 접지된 발의 미끄러짐까지 샷 텍스트·프레임 레이아웃을 가장 정확히 충족한다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S40sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김여상: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:990143>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "[Critical] 프롬프트가 명시한 '미디엄 샷(medium shot)'이 아닌 인물의 전신이 모두 보이는 풀 샷으로 렌더링되었으며, 프레임 상단 중앙에 있어야 할 산책로 가장자리가 좌측 배경으로 치우쳐 샷 스케일 및 레이아웃 지시를 위반했습니다.",
     "fix_en": "Since correcting the shot scale and background layout requires moving the camera, mitigate in place by refining the details of the character's screaming expression. Preserve the current full-body framing, the existing placement of the background slope and trees, his green rain gear, his mid-air falling pose, and the low-key nighttime lighting.",
     "needs_regeneration": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Since correcting the shot scale and background layout requires moving the camera, mitigate in place by refining the details of the character's screaming expression. Preserve the current full-body framing, the existing placement of the background slope and trees, his green rain gear, his mid-air falling pose, and the low-key nighttime lighting.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "중간 오른쪽의 추락 자세, 위쪽 길과 어두운 숲 위치는 충실하지만 입 벌림과 얼굴 긴장이 B보다 약해 핵심인 ‘비명을 지르듯 크게 벌린 입’이 덜 선명하며, 두 후보 모두 엄밀한 미디엄 숏보다 다소 넓다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "A의 장소·의상·자세·배경·빛을 그대로 유지하면서 입과 얼굴의 비명 연기만 강화해 핵심 순간을 더 정확히 고쳤고, 요청하지 않은 인물·의복·물체 변화는 보이지 않는다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.764,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.764,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "중간 오른쪽의 추락 자세, 위쪽 길과 어두운 숲 위치는 충실하지만 입 벌림과 얼굴 긴장이 B보다 약해 핵심인 ‘비명을 지르듯 크게 벌린 입’이 덜 선명하며, 두 후보 모두 엄밀한 미디엄 숏보다 다소 넓다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "A의 장소·의상·자세·배경·빛을 그대로 유지하면서 입과 얼굴의 비명 연기만 강화해 핵심 순간을 더 정확히 고쳤고, 요청하지 않은 인물·의복·물체 변화는 보이지 않는다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "미디엄샷 스케일·중앙우측 배치·상단 길 가장자리라는 프레임 지시를 더 정확히 지키고, 비명 표정과 실루엣이 저조도 속에서 가장 또렷하며 지면 접촉·낙하 궤적이 명확해 계약 이행도가 높다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "동일 구성이지만 계약이 요구하지 않은 변경(프레임 확대로 인물 축소, 길 가장자리 흙 벽면 확대와 지면 텍스처 단순화, 표정·손 디테일 약화)이 누적되어 '비명 지르는 표정을 실루엣으로 살린다'는 조명·연기 지시에서 손해를 봤다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.639
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.639
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "미디엄샷 스케일·중앙우측 배치·상단 길 가장자리라는 프레임 지시를 더 정확히 지키고, 비명 표정과 실루엣이 저조도 속에서 가장 또렷하며 지면 접촉·낙하 궤적이 명확해 계약 이행도가 높다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "동일 구성이지만 계약이 요구하지 않은 변경(프레임 확대로 인물 축소, 길 가장자리 흙 벽면 확대와 지면 텍스처 단순화, 표정·손 디테일 약화)이 누적되어 '비명 지르는 표정을 실루엣으로 살린다'는 조명·연기 지시에서 손해를 봤다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 13,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S40sh1"
  }
 },
 "S41sh3": {
  "input_fingerprint": "657711af7e7804fb",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 비석 앞에 놓인 썩은 육고기 덩어리의 선명한 모습.\n\nLOCATION (lock): A small forest clearing around an old moss-covered marker stone, with rotten meat placed at its base. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Near ground level and directly facing the marker, the final dolly-in settles on the rotten meat in the lower center while keeping the marker base visible behind it as positional evidence. The meat remains below forty percent of the frame, with the surrounding ground and lower portion of the marker maintaining realistic scale.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: rotten meat (Rotten and placed in front of the marker); used as Primary focal object in the lower center; marker (Old and heavily covered with moss) — Only its lower front face is visible behind the meat; no inscription is emphasized in this framing; used as Partial background context immediately behind the meat.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime ambient light appropriate to the forest reveals the rotten meat and moss-covered marker without introducing a new light source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The moss-covered marker stones remain standing with rotting meat placed before each one. Jihu’s necklace remains missing.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person appears in frame, and no face, and no figure standing, sitting or passing through. Do not invent someone to occupy the space.\n\nOne thing this rule does NOT forbid: the hand or arm of whoever is doing what the shot text describes. If the moment is an object being held, carried, operated, read, aimed, worn or handed over, the hand doing it belongs in frame, resting on or gripping that object exactly as its real use requires. An object photographed floating unsupported, with the hand that must be holding it erased, is a worse failure than the person you were avoiding — render the hand.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "C": 9,
   "A": 7,
   "B": 5
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 9,
    "verdict_ko": "지면 높이 정면 클로즈업, 하단 중앙의 썩은 육고기, 배경의 이끼 낀 비석 하부라는 카메라 지시를 모두 충족하고 레퍼런스 현장(기단·오솔길·관목)도 가장 정확히 재현했으며 고기 재질도 실제 부패육으로 읽힌다."
   },
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "구도와 접지·스케일은 규정대로지만 고기가 탈색된 지방 덩어리처럼 보여 '육고기' 질감이 약하고, 기단이 레퍼런스와 달리 이끼로 뒤덮여 구조 디테일이 어긋난다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "고기 배치는 맞으나 프레임이 과도하게 타이트해 숲 공터 맥락이 사라졌고, 상하 검은 띠(레터박스)와 비신에 인물처럼 보이는 얼룩이 생겨 장소 재현과 프레임 순수성이 가장 떨어진다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_path_network_sel.png"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "seed-bg만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_bypass:bg_only"
 },
 "S41sh5": {
  "input_fingerprint": "2d0aa78130cc26f1",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 멀리서 날아온 휘파람 소리 쪽 숲으로 고개를 돌린 채 의아하게 응시하는 김지후(강수리영)의 측면.\n\nLOCATION (lock): The wooded edge of the marker-stone clearing among dense black pines. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At head height beside 김지후(강수리영), the camera holds a close profile aimed across her sightline after her head has turned toward the whistle. She occupies the near left half with her breathing posture still tense, while densely packed trunks and undergrowth remain visible in the direction of her questioning stare.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: dense pine trunks (Densely packed) — Their vertical sides recede across the right side of the frame; used as Layered background aligned with 김지후(강수리영)'s searching gaze; undergrowth between the trees (The only visible cover between the trunks); used as Low visual barrier beneath her sightline.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime ambient light keeps her profile readable while retaining restrained contrast within the dense forest.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same moss-covered marker stone, rotten meat offering, dense pine trunks, undergrowth, and daylight. Exclude the meat as the sole focal subject; show the woman turning toward the whistle.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu remains without her necklace while following the whistle through the forest.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 9,
   "C": 5
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "머리높이 측면·우측 밀집 수간 배치와 시선 방향은 정확하나, 이전 스틸의 표석과 고기 제물이 전혀 보이지 않아 장소 잠금 근거가 약하다."
   },
   {
    "label": "B",
    "score": 9,
    "verdict_ko": "측면 클로즈업, 우측으로 뻗은 응시 축, 층층이 겹친 소나무 줄기와 하부 덤불, 그리고 표석·썩은 고기까지 유지해 샷 텍스트와 장소 잠금을 모두 충족한다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "'측면으로 숲을 응시'가 아니라 카메라 쪽으로 돌아본 후면 3/4이어서 시선 방향과 지정 구도가 어긋난다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S41sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:813104>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "화면 하단 모서리에 레퍼런스에 있던 썩은 고깃덩어리 대신 주름진 크러스트가 있는 체리 파이 형태의 음식이 놓여 있습니다.",
     "fix_en": "Replace the pie-like object at the bottom center edge of the frame with a pile of raw, glistening, decaying meat and fat matching the reference still, while strictly preserving the woman, her pose, her clothing, the stone pillar behind her, the dense pine forest, the lighting, and the existing framing.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "캐릭터 레퍼런스에서 착용하고 있는 회색 캡 모자를 쓰지 않고 맨머리로 렌더링되었습니다.",
     "fix_en": "",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "인물 뒤에 보이는 비석이 레퍼런스의 넓은 기단부가 있는 형태가 아니라 매끄러운 단일 기둥이며, 이끼의 형태도 질감이 다르고 지나치게 밝은 연두색으로 잘못 표현되었습니다.",
     "fix_en": "",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the pie-like object at the bottom center edge of the frame with a pile of raw, glistening, decaying meat and fat matching the reference still, while strictly preserving the woman, her pose, her clothing, the stone pillar behind her, the dense pine forest, the lighting, and the existing framing.\n- \n- \n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 프레임 좌측 근경에 측면(우향)으로 서 있고, 시선과 얼굴이 화면 오른쪽 소나무 숲 깊은 곳을 향한다. 카메라는 머리 높이에서 그녀의 시선축을 가로질러 근접 측면을 잡았고, 오른쪽 절반에 수직 소나무 줄기들과 하부 덤불이 그녀의 시선 방향으로 층층이 배치되어 있다. 휘파람이 날아온 방향을 향한 응시가 실제로 숲 속에 착지한다.",
      "built_space": "인공 구조물은 이끼 낀 표석뿐. 그녀 뒤(좌측 중경)에 회청색 표석 상단과 이끼 낀 기단이 한 개만 보이며, 이전 스틸의 재질·이끼 분포와 일치한다. 중복된 표석이나 추가 구조물 없음. 지면은 솔잎 깔린 흙길로 레퍼런스와 동일.",
      "entities": "여성은 20대 초 한국인, 긴 검은 생머리, 갸름한 얼굴로 캐릭터 레퍼런스와 부합. 어두운 갈색 워크재킷 착용, 목걸이(뿔 펜던트) 없음 — CARRIED STATE 준수. 모자는 프레임 밖/미착용이나 클로즈업 범위상 감점 대상 아님. 썩은 고기는 좌측 하단 구석에 작게 일부만 보여 '주된 초점 제외' 지시에 부합하고 실제 크기감도 유지. 눈은 정상 홍채·공막.",
      "hard_violations": [],
      "physics": "그녀는 상체를 앞으로 약간 숙인 채 두 발로 지면에 서 있는(허리 굽힘) 자세로, 어깨·재킷의 주름과 중력 방향이 일치한다. 공중에 떠 있는 물체 없음. 고기는 지면에 놓여 있고, 나무는 뿌리로 지면에 접해 있다."
     },
     {
      "label": "B",
      "direction": "A와 동일하게 김지후는 좌측 근경 측면, 시선은 우측 숲 속을 향한다. 다만 화면 하단 중앙에 거대하게 배치된 고깃덩이가 시선축 아래를 가로막아 시각적 초점을 강탈한다. 시선 자체는 숲에 착지.",
      "built_space": "표석은 여전히 좌측 중경에 하나. 그러나 표석 기단 앞의 고기가 표석보다 훨씬 크게 렌더되어, 실제 물리적 스케일 관계가 붕괴했다(중경의 고기가 근경 인물 얼굴만 한 크기로 확대).",
      "entities": "인물 동일: 20대 한국 여성, 긴 검은 생머리, 어두운 재킷, 목걸이 없음 — 레퍼런스 부합. 썩은 고기는 이전 스틸의 구더기·부패 질감과 일치하나 크기가 비현실적으로 확대되어 전경 존재감으로 승격됨 — 'Exclude the meat as the sole focal subject' 및 '배경 물체를 전경 존재로 확대하지 말 것' 지시 위반.",
      "hard_violations": [
       "중경 지면의 고깃덩이가 실제 크기·거리 대비 과도하게 확대되어 전경 오브젝트로 부유하듯 배치됨 — 명시된 '진짜 물리적 스케일 유지' 규정 위반이며 지면 접지 근거가 불명확"
      ],
      "physics": "인물은 A와 동일하게 두 발로 서서 상체를 숙인 자세로 지지된다. 그러나 고기는 지면 접지선이 프레임 하단에 잘려 보이지 않고 크기가 주변 표석·덤불과 맞지 않아, 지면에 놓인 것이 아니라 렌즈 앞에 떠 있는 것처럼 읽힌다. 이를 지지하는 표면이 확인되지 않는다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "머리 높이 근접 측면, 좌측 근경 인물·우측 층층 소나무 줄기와 하부 덤불이라는 카메라 지시를 정확히 충족하고, 고기는 구석에 작게 남아 '주 초점 제외' 조건도 지킨다. 목걸이 미착용·인물 동일성·주간 자연광 모두 부합."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "인물·시선·배경은 A와 사실상 동일하나, 고깃덩이를 화면 하단 중앙에 비현실적 크기로 확대해 전경 주체로 만든 변경이 '고기를 단독 초점에서 제외' 및 '배경 물체를 전경으로 확대 금지' 조항을 정면으로 위반하고, 접지 근거도 없어 떠 보인다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "머리 높이 근접 측면, 좌측 근경 인물·우측 층층 소나무 줄기와 하부 덤불이라는 카메라 지시를 정확히 충족하고, 고기는 구석에 작게 남아 '주 초점 제외' 조건도 지킨다. 목걸이 미착용·인물 동일성·주간 자연광 모두 부합."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "인물·시선·배경은 A와 사실상 동일하나, 고깃덩이를 화면 하단 중앙에 비현실적 크기로 확대해 전경 주체로 만든 변경이 '고기를 단독 초점에서 제외' 및 '배경 물체를 전경으로 확대 금지' 조항을 정면으로 위반하고, 접지 근거도 없어 떠 보인다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 프레임 좌측 근경에 측면(우향)으로 서 있고, 시선과 얼굴이 화면 오른쪽 소나무 숲 깊은 곳을 향한다. 카메라는 머리 높이에서 그녀의 시선축을 가로질러 근접 측면을 잡았고, 오른쪽 절반에 수직 소나무 줄기들과 하부 덤불이 그녀의 시선 방향으로 층층이 배치되어 있다. 휘파람이 날아온 방향을 향한 응시가 실제로 숲 속에 착지한다.",
      "built_space": "인공 구조물은 이끼 낀 표석뿐. 그녀 뒤(좌측 중경)에 회청색 표석 상단과 이끼 낀 기단이 한 개만 보이며, 이전 스틸의 재질·이끼 분포와 일치한다. 중복된 표석이나 추가 구조물 없음. 지면은 솔잎 깔린 흙길로 레퍼런스와 동일.",
      "entities": "여성은 20대 초 한국인, 긴 검은 생머리, 갸름한 얼굴로 캐릭터 레퍼런스와 부합. 어두운 갈색 워크재킷 착용, 목걸이(뿔 펜던트) 없음 — CARRIED STATE 준수. 모자는 프레임 밖/미착용이나 클로즈업 범위상 감점 대상 아님. 썩은 고기는 좌측 하단 구석에 작게 일부만 보여 '주된 초점 제외' 지시에 부합하고 실제 크기감도 유지. 눈은 정상 홍채·공막.",
      "hard_violations": [],
      "physics": "그녀는 상체를 앞으로 약간 숙인 채 두 발로 지면에 서 있는(허리 굽힘) 자세로, 어깨·재킷의 주름과 중력 방향이 일치한다. 공중에 떠 있는 물체 없음. 고기는 지면에 놓여 있고, 나무는 뿌리로 지면에 접해 있다."
     },
     {
      "label": "B",
      "direction": "A와 동일하게 김지후는 좌측 근경 측면, 시선은 우측 숲 속을 향한다. 다만 화면 하단 중앙에 거대하게 배치된 고깃덩이가 시선축 아래를 가로막아 시각적 초점을 강탈한다. 시선 자체는 숲에 착지.",
      "built_space": "표석은 여전히 좌측 중경에 하나. 그러나 표석 기단 앞의 고기가 표석보다 훨씬 크게 렌더되어, 실제 물리적 스케일 관계가 붕괴했다(중경의 고기가 근경 인물 얼굴만 한 크기로 확대).",
      "entities": "인물 동일: 20대 한국 여성, 긴 검은 생머리, 어두운 재킷, 목걸이 없음 — 레퍼런스 부합. 썩은 고기는 이전 스틸의 구더기·부패 질감과 일치하나 크기가 비현실적으로 확대되어 전경 존재감으로 승격됨 — 'Exclude the meat as the sole focal subject' 및 '배경 물체를 전경 존재로 확대하지 말 것' 지시 위반.",
      "hard_violations": [
       "중경 지면의 고깃덩이가 실제 크기·거리 대비 과도하게 확대되어 전경 오브젝트로 부유하듯 배치됨 — 명시된 '진짜 물리적 스케일 유지' 규정 위반이며 지면 접지 근거가 불명확"
      ],
      "physics": "인물은 A와 동일하게 두 발로 서서 상체를 숙인 자세로 지지된다. 그러나 고기는 지면 접지선이 프레임 하단에 잘려 보이지 않고 크기가 주변 표석·덤불과 맞지 않아, 지면에 놓인 것이 아니라 렌즈 앞에 떠 있는 것처럼 읽힌다. 이를 지지하는 표면이 확인되지 않는다."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 화면 좌측 근경에 측면으로 서 있고, 시선은 프레임 오른쪽(소나무 줄기가 겹쳐진 숲 깊은 곳)을 향한다. 고개가 휘파람 소리 방향으로 돌아간 상태로 읽히며, 시선 축과 배경의 나무 기둥 층이 일치한다.",
      "built_space": "인공 구조물은 이끼 낀 표석뿐. 여성 뒤쪽 중경에 회녹색 표석 상단과 이끼 낀 기단이 보이며, 이전 스틸의 재질·형태와 일치한다. 표석은 하나만 존재해 개수 오류 없음.",
      "entities": "인물은 20대 초반 한국 여성, 긴 검은 생머리, 참조와 유사한 얼굴형. 목걸이 없음(CARRIED STATE 충족). 모자는 없으나 프레임상 머리 위 여백이 있어 모자 부재가 눈에 띈다. 문제는 우측 하단의 썩은 고기 덩어리로, 이전 스틸에서 지면에 놓였던 크기가 여기서는 인물 얼굴만 한 전경 덩어리로 확대되어 표석 기단보다도 크게 렌더링됐다 — 실제 물리적 스케일 위반이자 '고기를 단독 초점 피사체로 배제' 지시와 충돌.",
      "hard_violations": [
       "배경/지면 소품인 썩은 고기가 실제 크기를 무시하고 전경 거대 오브젝트로 확대되어 프레임 하단을 점유 — '배경 사물을 전경 존재로 확대하지 말 것' 및 참 물리 스케일 조항 위반",
       "그 고기 덩어리는 지면 접지나 받침이 보이지 않고 프레임 하단에 떠 있는 듯 배치되어 지지물이 확인되지 않음"
      ],
      "physics": "여성은 상체를 앞으로 기울인 자세로 하반신은 프레임 밖이나 몸통 기울기와 어깨 긴장이 자연스럽다. 반면 고기 덩어리는 표석 기단 앞 지면보다 훨씬 앞·위에 떠 있는 위치에 있어 무엇이 받치는지 보이지 않는다."
     },
     {
      "label": "B",
      "direction": "동일하게 좌측 근경 측면, 시선은 우측 숲 안쪽 나무 기둥들 사이를 향한다. 고개 회전 방향과 시선 축이 배경의 층진 소나무 기둥과 정확히 정렬되고, 시선 아래로 관목 덤불이 낮은 시각적 장벽으로 깔린다.",
      "built_space": "표석 하나가 인물 뒤 중경에 이끼 낀 기단과 함께 서 있고 크기·거리 비율이 실제 스케일에 맞다. 중복 구조물 없음.",
      "entities": "20대 초반 한국 여성, 긴 검은 생머리, 참조 얼굴과 일치. 갈색 워크재킷 동일, 목걸이 없음. 썩은 고기는 좌측 하단 지면에 작게 일부만 보여 이전 스틸의 소품이 유지되되 단독 초점이 아니다.",
      "hard_violations": [],
      "physics": "여성은 상체를 앞으로 숙인 채 어깨·목에 긴장이 남아 있고, 프레임 밖 하반신이 체중을 받치는 것으로 읽힌다. 고기 덩어리는 지면에 놓여 접지되어 있다. 공중에 뜬 물체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "인물 연기와 시선 방향은 맞지만, 지면 소품인 썩은 고기를 얼굴 크기의 전경 덩어리로 확대해 프레임 하단을 점유시켜 '실제 스케일 유지·고기를 단독 초점에서 배제' 조항을 정면으로 어겼고 지지면도 불명확하다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "근접 측면·좌측 반쪽 배치·우측으로 층진 소나무 기둥과 덤불이라는 카메라 지시를 그대로 지키고, 고기는 지면에 작게 남겨 스케일과 '초점 배제' 지시를 모두 충족 — 인물 동일성·의상·목걸이 부재도 유지."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "인물 연기와 시선 방향은 맞지만, 지면 소품인 썩은 고기를 얼굴 크기의 전경 덩어리로 확대해 프레임 하단을 점유시켜 '실제 스케일 유지·고기를 단독 초점에서 배제' 조항을 정면으로 어겼고 지지면도 불명확하다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "근접 측면·좌측 반쪽 배치·우측으로 층진 소나무 기둥과 덤불이라는 카메라 지시를 그대로 지키고, 고기는 지면에 작게 남겨 스케일과 '초점 배제' 지시를 모두 충족 — 인물 동일성·의상·목걸이 부재도 유지."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "김지후는 화면 좌측 근경에 측면으로 서 있고, 시선은 프레임 오른쪽(소나무 줄기가 겹쳐진 숲 깊은 곳)을 향한다. 고개가 휘파람 소리 방향으로 돌아간 상태로 읽히며, 시선 축과 배경의 나무 기둥 층이 일치한다.",
      "built_space": "인공 구조물은 이끼 낀 표석뿐. 여성 뒤쪽 중경에 회녹색 표석 상단과 이끼 낀 기단이 보이며, 이전 스틸의 재질·형태와 일치한다. 표석은 하나만 존재해 개수 오류 없음.",
      "entities": "인물은 20대 초반 한국 여성, 긴 검은 생머리, 참조와 유사한 얼굴형. 목걸이 없음(CARRIED STATE 충족). 모자는 없으나 프레임상 머리 위 여백이 있어 모자 부재가 눈에 띈다. 문제는 우측 하단의 썩은 고기 덩어리로, 이전 스틸에서 지면에 놓였던 크기가 여기서는 인물 얼굴만 한 전경 덩어리로 확대되어 표석 기단보다도 크게 렌더링됐다 — 실제 물리적 스케일 위반이자 '고기를 단독 초점 피사체로 배제' 지시와 충돌.",
      "hard_violations": [
       "배경/지면 소품인 썩은 고기가 실제 크기를 무시하고 전경 거대 오브젝트로 확대되어 프레임 하단을 점유 — '배경 사물을 전경 존재로 확대하지 말 것' 및 참 물리 스케일 조항 위반",
       "그 고기 덩어리는 지면 접지나 받침이 보이지 않고 프레임 하단에 떠 있는 듯 배치되어 지지물이 확인되지 않음"
      ],
      "physics": "여성은 상체를 앞으로 기울인 자세로 하반신은 프레임 밖이나 몸통 기울기와 어깨 긴장이 자연스럽다. 반면 고기 덩어리는 표석 기단 앞 지면보다 훨씬 앞·위에 떠 있는 위치에 있어 무엇이 받치는지 보이지 않는다."
     },
     {
      "label": "A",
      "direction": "동일하게 좌측 근경 측면, 시선은 우측 숲 안쪽 나무 기둥들 사이를 향한다. 고개 회전 방향과 시선 축이 배경의 층진 소나무 기둥과 정확히 정렬되고, 시선 아래로 관목 덤불이 낮은 시각적 장벽으로 깔린다.",
      "built_space": "표석 하나가 인물 뒤 중경에 이끼 낀 기단과 함께 서 있고 크기·거리 비율이 실제 스케일에 맞다. 중복 구조물 없음.",
      "entities": "20대 초반 한국 여성, 긴 검은 생머리, 참조 얼굴과 일치. 갈색 워크재킷 동일, 목걸이 없음. 썩은 고기는 좌측 하단 지면에 작게 일부만 보여 이전 스틸의 소품이 유지되되 단독 초점이 아니다.",
      "hard_violations": [],
      "physics": "여성은 상체를 앞으로 숙인 채 어깨·목에 긴장이 남아 있고, 프레임 밖 하반신이 체중을 받치는 것으로 읽힌다. 고기 덩어리는 지면에 놓여 접지되어 있다. 공중에 뜬 물체 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 8
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S41sh3"
  }
 },
 "S44sh7": {
  "input_fingerprint": "0ddb2b39d58deb06",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day, direct sunlight.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)이 등 뒤에서 서늘하게 빛나는 도축용 칼을 빼어 든 순간.\n\nLOCATION (lock): The tool-lined work area inside the daylight slaughterhouse, directly behind the kneeling village youths. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track at 강영란(흰 머리 여자)'s waist-to-shoulder height from roughly behind and slightly off her right side, tightening on her back and drawing hand as the slaughter knife clears concealment. Her torso occupies the left half while the entire blade extends diagonally across the center-right, leaving the blood-marked slaughterhouse surfaces and hanging hooks as restrained depth cues.\n- FRAMING SCALE: medium shot\n- KEY BACKGROUND ELEMENTS: 깨진 창 (broken) — The broken window plane is angled behind 강영란(흰 머리 여자), with daylight entering through its open gaps; the exterior remains unresolved; used as The broken opening provides background depth behind the weapon reveal; 바닥과 벽의 핏자국 (marked with visible bloodstains); used as Visible beneath and beyond the figure as contextual evidence of the slaughterhouse; 고기 걸이 금속 걸개 (hanging and reflecting daylight) — Their curved sides hang behind and to the side of 강영란(흰 머리 여자), catching the camera obliquely; used as Narrow background highlights echo the revealed blade without competing with it; 도축용 칼 (drawn and reflecting daylight) — The blade is shown broadside to camera as it emerges behind 강영란(흰 머리 여자), with its full length visible; used as Primary weapon held fully within the composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Hard midday sunlight entering through the broken window creates low-key contrast, cold highlights on metal, and restrained blue-black, earthy, and deep-red tones.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same sunlit broken window, bloodstained floor and walls, hanging slaughter tools, and harsh daytime atmosphere. Exclude an empty establishing composition; add the elder drawing the slaughter knife behind the kneeling youths.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The bloodstained slaughterhouse and its fixed butchering tools remain around Youngran as she draws the knife.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 강영란(흰 머리 여자) right now, so 강영란(흰 머리 여자)'s hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 강영란(흰 머리 여자): its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (육혈귀 각성 상태) (Korean woman in her late 60s, neatly twisted white hair, graceful mature face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "카메라는 강영란의 오른쪽 뒤쪽 어깨 높이에서 등을 잡고 있고, 그녀의 시선은 화면 안쪽(무릎 꿇은 소년들 쪽)을 향한다. 칼끝은 화면 중앙-우측으로 대각선을 그리며 창과 무릎 꿇은 소년들 방향을 향한다. 칼날 광폭면(브로드사이드)이 카메라를 향해 전체 길이가 보인다 — 샷 텍스트 요구와 일치.",
    "built_space": "도축장 내부 구조가 레퍼런스와 정확히 일치: 좌측 페그보드와 걸린 칼들, 깨진 창(같은 파손 형상), 천장 레일과 갈고리 배열, 우측 문간과 벽 배전함, 안쪽 작업대. 갈고리 개수·위치가 레퍼런스와 대응하며 크기 과장 없음. 다만 상단 갈고리 하나가 그녀 머리 바로 뒤에 겹친다.",
    "entities": "흰(회백)머리를 단정히 틀어올린 60대 후반 한국 여성, 남색 자카드 한복풍 상의 — 캐릭터 레퍼런스와 일치. 손은 주름·검버섯·노년기 피부로 인물과 일치. 칼은 프롭 레퍼런스 중 세 번째(문양 새겨진 나무·황동 손잡이 칼)를 그대로 재현. 다만 화면 내 칼 크기가 실제 대비 다소 과장되어 보인다. 배경의 무릎 꿇은 소년 4명은 이전 컷 서사(무릎 꿇은 마을 소년들 뒤)와 로케이션 설명이 명시한 존재로, 후경 소품처럼 작게 배치.",
    "hard_violations": [],
    "physics": "칼은 오른손이 손잡이를 감싸 쥐고 등 뒤에서 뽑혀 나오는 상태 — 지지점 명확. 손목·전완이 소매에서 이어져 있고 힘줄과 무게감이 보인다. 소년들은 바닥에 무릎을 대고 접지되어 있음. 공중 부양 물체 없음."
   },
   {
    "label": "B",
    "direction": "카메라는 인물 정후방 정면(오른쪽 오프셋 거의 없음)에서 등을 잡는다. 칼은 몸 뒤에서 오른쪽 위로 대각선을 그려 방향성은 성립. 인물 시선은 보이지 않고, 무릎 꿇은 소년들은 좌측 하단 프레임 가장자리에 잘려 거의 보이지 않아 '무릎 꿇은 소년들 바로 뒤' 관계가 약하다.",
    "built_space": "레퍼런스 로케이션 재현은 양호(깨진 창, 페그보드, 갈고리 레일, 핏자국 바닥). 다만 우측 작업대가 레퍼런스보다 앞으로 당겨져 커졌고, 그 위에 레퍼런스에 없는 칼·도구들이 새로 배치되었다. 갈고리 배열은 대체로 일치.",
    "entities": "흰머리 쪽진 노년 여성 뒷모습, 남색 상의 — 인물 레퍼런스와 부합. 그러나 프레이밍이 풀샷에 가까워 '등과 칼 쥔 손을 타이트하게' 라는 미디엄샷 지시보다 넓다. 칼은 프롭 레퍼런스 어느 것과도 다른 거대한 검(마체테/시미터급) 형태로, 도축용 칼 레퍼런스 불일치가 크다. 손은 소매에 가려 손등 일부만 보이고 노년의 디테일이 약하다.",
    "hard_violations": [],
    "physics": "양손 중 오른손이 손잡이를 쥔 것으로 보이나 손 형태가 뭉개져 있고, 칼 무게(1m급 대형 검)에 비해 손목 자세가 부자연스럽다. 그래도 손이 손잡이에 닿아 있어 지지 자체는 존재. 바닥 접지 정상."
   },
   {
    "label": "C",
    "direction": "카메라는 정후방, 인물 전신 뒷모습. 칼은 오른쪽 앞으로 뻗어 대각선을 이루지만 칼끝이 위쪽 허공을 향한다. 왼팔은 등 뒤로 돌려 손을 보이나 칼과 무관.",
    "built_space": "창·페그보드·갈고리·핏자국 등 로케이션 재현은 무난. 다만 안쪽 작업대 다리 형태가 흐릿하고, 무릎 꿇은 소년들이 프레임에 전혀 없어 로케이션 락('무릎 꿇은 소년들 바로 뒤')이 성립하지 않는다.",
    "entities": "흰머리 쪽진 노년 여성, 남색 상의 — 인물 레퍼런스 부합. 칼은 프롭 레퍼런스에 없는 대형 보위/시미터형 나이프로 불일치, 크기도 과장. 칼을 쥔 오른손은 손잡이 위치와 어긋나 손가락이 손잡이를 감싸지 않고 칼날 밑동 근처에서 뭉개져 있으며, 손잡이 끝이 손 아래로 떠 있는 것처럼 보인다.",
    "hard_violations": [
     "오른손이 손잡이를 실제로 감싸 쥐지 않은 채 대형 칼이 손 앞쪽에 걸쳐 있어, 칼을 지지하는 그립이 해부학적으로 성립하지 않음(무게 지지점 부재)"
    ],
    "physics": "칼 손잡이와 손의 접점이 불명확 — 손가락 관절이 손잡이를 감싸지 않고, 손잡이 하단이 손 밖으로 떨어져 나온 상태로 칼이 사실상 지지 없이 떠 보인다. 하체는 접지 정상."
   }
  ],
  "totals": {
   "A": 9,
   "B": 5,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 9,
    "verdict_ko": "어깨 높이·우측 후방 트랙, 좌반부의 등과 중앙우측을 가로지르는 칼날 전체 노출까지 카메라 지시를 가장 정확히 이행했고, 프롭 레퍼런스의 문양 칼과 노년의 손, 무릎 꿇은 소년들 뒤라는 로케이션 락까지 유일하게 충족했다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "로케이션 재현은 좋으나 프레임이 미디엄샷보다 넓고 정후방이며, 칼이 레퍼런스에 없는 1m급 대형 검으로 '도축용 칼' 프롭 락을 크게 벗어났고 작업대 위 도구가 새로 추가되었다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "무릎 꿇은 소년들이 프레임에서 사라져 로케이션 관계가 무너졌고, 칼 역시 레퍼런스와 다른 대형 나이프인 데다 손이 손잡이를 쥐지 못해 무기의 지지점 자체가 성립하지 않는다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S44sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자) (육혈귀 각성 상태): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:855048>"
   },
   {
    "label": "PROP REFERENCE — 칼(도축용 칼·신칼 포함): the exact object appearing in this shot; match its look, material and wear exactly.",
    "path": "<bytes:1220370>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "칼을 쥔 팔이 몸과 분리되어 등 위에 떠 있는 것처럼 기형적으로 묘사되었습니다.",
     "fix_en": "Redraw the hand and sleeve holding the knife to connect organically to the woman's right side as her right arm, preserving her pose, clothing, the knife, background figures, room, and lighting.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "여성의 머리가 레퍼런스의 단정한 올림머리와 달리 땋은 머리로 잘못 묘사되었습니다.",
     "fix_en": "",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the hand and sleeve holding the knife to connect organically to the woman's right side as her right arm, preserving her pose, clothing, the knife, background figures, room, and lighting.\n- \n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "구도·발도 동작·인물/칼 일치는 충족하나, 뒤통수와 목덜미가 뭉개지고 창가에 과한 흐림·먼지 번짐이 생겨 재질감과 인물 식별성이 B보다 떨어진다.  ★위반: [gpt] 강영란 외에 무릎 꿇은 인물 네 명을 추가해, 화면에 등장할 수 있는 사람을 강영란으로만 제한한 계약을 위반했다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일 구도와 등 뒤 발도 순간을 유지하면서 백발 쪽머리·목덜미 디테일과 깨진 창·금속 하이라이트의 물성이 더 정확해, 계약이 요구한 인물 일치와 재질 사실성에서 앞선다.  ★위반: [gpt] 강영란 외에 무릎 꿇은 인물 네 명을 추가해, 화면에 등장할 수 있는 사람을 강영란으로만 제한한 계약을 위반했다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.875,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.625,
      "B": 1.75
     },
     "violations": {
      "A": [
       "[gpt] 강영란 외에 무릎 꿇은 인물 네 명을 추가해, 화면에 등장할 수 있는 사람을 강영란으로만 제한한 계약을 위반했다."
      ],
      "B": [
       "[gpt] 강영란 외에 무릎 꿇은 인물 네 명을 추가해, 화면에 등장할 수 있는 사람을 강영란으로만 제한한 계약을 위반했다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "A"
     },
     "agreed": false
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "구도·발도 동작·인물/칼 일치는 충족하나, 뒤통수와 목덜미가 뭉개지고 창가에 과한 흐림·먼지 번짐이 생겨 재질감과 인물 식별성이 B보다 떨어진다.  ★위반: [gpt] 강영란 외에 무릎 꿇은 인물 네 명을 추가해, 화면에 등장할 수 있는 사람을 강영란으로만 제한한 계약을 위반했다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일 구도와 등 뒤 발도 순간을 유지하면서 백발 쪽머리·목덜미 디테일과 깨진 창·금속 하이라이트의 물성이 더 정확해, 계약이 요구한 인물 일치와 재질 사실성에서 앞선다.  ★위반: [gpt] 강영란 외에 무릎 꿇은 인물 네 명을 추가해, 화면에 등장할 수 있는 사람을 강영란으로만 제한한 계약을 위반했다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지정된 뒤·우측 미디엄샷, 좌반부 등·우측 대각 칼날, 깨진 창의 파편 형태와 바닥 핏자국·갈고리 등 이전 스틸 고정 요소를 가장 선명하게 보존했고 칼·인물 레퍼런스 일치도 높다.  ★위반: [gpt] 강영란만 등장할 수 있다는 PEOPLE 제한에도 불구하고 무릎 꿇은 남성 네 명이 추가로 화면에 등장한다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "구도·인물·칼은 A와 거의 동일하나 깨진 창의 개구부가 하얗게 뭉개져 핵심 배경 요소인 '깨진 창'과 저조도 대비·핏자국 증거성이 약화됐고, 소년 실루엣도 더 뭉개져 계약이 요구하지 않은 손실이 발생했다.  ★위반: [gpt] 강영란만 등장할 수 있다는 PEOPLE 제한에도 불구하고 무릎 꿇은 남성 네 명이 추가로 화면에 등장한다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.75
     },
     "adjusted": {
      "A": 1.75,
      "B": 1.5
     },
     "violations": {
      "A": [
       "[gpt] 강영란만 등장할 수 있다는 PEOPLE 제한에도 불구하고 무릎 꿇은 남성 네 명이 추가로 화면에 등장한다."
      ],
      "B": [
       "[gpt] 강영란만 등장할 수 있다는 PEOPLE 제한에도 불구하고 무릎 꿇은 남성 네 명이 추가로 화면에 등장한다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 뒤·우측 미디엄샷, 좌반부 등·우측 대각 칼날, 깨진 창의 파편 형태와 바닥 핏자국·갈고리 등 이전 스틸 고정 요소를 가장 선명하게 보존했고 칼·인물 레퍼런스 일치도 높다.  ★위반: [gpt] 강영란만 등장할 수 있다는 PEOPLE 제한에도 불구하고 무릎 꿇은 남성 네 명이 추가로 화면에 등장한다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "구도·인물·칼은 A와 거의 동일하나 깨진 창의 개구부가 하얗게 뭉개져 핵심 배경 요소인 '깨진 창'과 저조도 대비·핏자국 증거성이 약화됐고, 소년 실루엣도 더 뭉개져 계약이 요구하지 않은 손실이 발생했다.  ★위반: [gpt] 강영란만 등장할 수 있다는 PEOPLE 제한에도 불구하고 무릎 꿇은 남성 네 명이 추가로 화면에 등장한다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 13,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S44sh1"
  }
 },
 "S45sh9": {
  "input_fingerprint": "24ff928a90194ba6",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 바닥에 무릎을 꿇고 앉아 흐느끼는 고산들의 머리 위에 손을 얹은 채 머무는 강영란(흰 머리 여자)의 자세.\n\nLOCATION (lock): The ground of the secluded black-pine trail where the two kneel together. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Descend to slightly above kneeling eye level at a three-quarter side angle, framing both characters' complete kneeling postures at medium distance. 강영란(흰 머리 여자) occupies the left side with her hand visibly resting on 고산들's bowed head, while his folded body anchors the lower right and the path remains visible around them.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the middle-left of the frame, midground, reaches for 고산들's bowed head; 고산들 in the lower-right of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 숲길 (present); used as Provides visible ground around both kneeling bodies and clarifies their lowered level; 숲의 나무들 (present) — The trunks appear obliquely behind the kneeling pair; used as Forms restrained vertical depth behind the consoling gesture.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural forest daylight retains solemn contrast with muted earth tones, aged ivory, and dense blue-black separation.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same daylight pine path, filtered forest light, ground texture, and the pair's clothing. Exclude their earlier standing positions; show the man kneeling while the elder rests a hand on his head.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (Korean woman in her late 60s, neatly twisted white hair, graceful dignified face); 고산들 (Korean man in his mid-40s, closely cropped dark hair, cold angular face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 6,
   "C": 7
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "3/4 측면·무릎 눈높이 약간 위 앵글과 좌중간/우하단 배치, 머리에 얹은 손을 모두 충족하고 두 인물의 의상(올리브 조끼·체크셔츠, 긴 손톱 포함)까지 이전 스틸·캐릭터 참조와 가장 정확히 일치한다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "장소 재현(울타리·건물·침목)은 가장 충실하나 카메라가 거의 정측면이고, 남자의 올리브 조끼·파우치와 여자의 노리개·긴 손톱이 사라져 '이전 스틸의 복장 유지' 조항을 어겼다."
   },
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "자세·손 위치·숲길 배경은 적절하나 남자의 올리브 조끼와 여자의 긴 손톱 등 참조 복장 요소가 빠지고 앵글이 3/4보다 정측면에 가깝다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S45sh3_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:1013203>"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903959>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "남자의 머리에 얹은 여자의 팔이 어깨가 아닌 가슴 정면(옷고름 위)에서 기형적으로 뻗어 나와 있습니다.",
     "fix_en": "Redraw the woman's extended arm so it originates naturally from her right shoulder behind her torso, removing the anomalous arm from her front collar, while keeping her face, her left arm on her lap, the kneeling man, their clothing, and the forest background exactly as they are.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the woman's extended arm so it originates naturally from her right shoulder behind her torso, removing the anomalous arm from her front collar, while keeping her face, her left arm on her lap, the kneeling man, their clothing, and the forest background exactly as they are.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "무릎 꿇은 두 사람의 전신 자세, 좌측 중간 강영란·우측 하단 고산들 배치, 머리 위 손 접촉, 이전 스틸의 숲길·의상·빛이 모두 충족되며 팔 소매의 드레이프와 접촉점이 물리적으로 자연스럽다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "구도·장소·의상은 A와 동일하게 계약을 충족하지만, 계약이 요구하지 않은 변경(뻗은 팔 소매 부피 축소로 팔이 가늘어지고 손-머리 접촉이 얕아짐, 배경 잎사귀 정리)이 있어 재질감과 접촉의 설득력에서 손해를 봤다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.875
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.875
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "무릎 꿇은 두 사람의 전신 자세, 좌측 중간 강영란·우측 하단 고산들 배치, 머리 위 손 접촉, 이전 스틸의 숲길·의상·빛이 모두 충족되며 팔 소매의 드레이프와 접촉점이 물리적으로 자연스럽다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "구도·장소·의상은 A와 동일하게 계약을 충족하지만, 계약이 요구하지 않은 변경(뻗은 팔 소매 부피 축소로 팔이 가늘어지고 손-머리 접촉이 얕아짐, 배경 잎사귀 정리)이 있어 재질감과 접촉의 설득력에서 손해를 봤다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "접촉·자세·의상은 동일하게 맞지만 프레임이 더 타이트해 '와이드 샷·경로가 주변에 보이도록'이라는 프레이밍 요구를 덜 충족하고, 인물 크기가 커져 미디엄에 가깝다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 접촉·정체성·의상을 유지하면서 두 사람의 완전한 무릎 자세와 주변 숲길·나무 줄기 깊이를 더 넓게 담아 지정된 와이드 샷·프레임 배치(좌중앙 강영란, 우하단 고산들)를 더 정확히 실현한다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.764,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.764,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "접촉·자세·의상은 동일하게 맞지만 프레임이 더 타이트해 '와이드 샷·경로가 주변에 보이도록'이라는 프레이밍 요구를 덜 충족하고, 인물 크기가 커져 미디엄에 가깝다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "동일한 접촉·정체성·의상을 유지하면서 두 사람의 완전한 무릎 자세와 주변 숲길·나무 줄기 깊이를 더 넓게 담아 지정된 와이드 샷·프레임 배치(좌중앙 강영란, 우하단 고산들)를 더 정확히 실현한다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 14
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S45sh3"
  }
 },
 "S46sh2::bgfirst_bg": {
  "input_fingerprint": "adb77b8abcb2eb1f",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 김지후(강수리영)가 낡은 비석 표면에 선명하게 음각된 '수 리 영' 세 글자를 내려다보는 시선.\n\nLOCATION (lock): The marker-stone clearing deep in the forest, among several stones all bearing the same name.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Settle just behind 김지후(강수리영)'s shoulder at close range, using her shoulder and lowered head as a near-edge frame for the marker. The inscribed face occupies the central lower portion without exceeding a natural scale, and the three characters “수 리 영” sit precisely at the endpoint of her downward eyeline.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 비석의 ‘수 리 영’ 음각면 in the lower-center of the frame, midground; 김지후(강수리영) in the middle-left of the frame, foreground, looks toward inscribed marker face.\n- KEY BACKGROUND ELEMENTS: 낡은 비석 (old, with the name clearly incised) — Its inscribed front face is angled toward camera, clearly showing the recessed Korean characters ‘수 리 영’; used as Primary focal surface at the end of 김지후(강수리영)'s gaze; 주변의 다른 비석들 (present, each bearing the same name) — Only partial side and front planes are visible farther into the forest; used as Provides limited depth around the marker and shoulder framing.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime forest light renders the inscription clearly while preserving ominous high contrast and subdued earthy, aged-ivory, and blue-black tones.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 김지후(강수리영)가 낡은 비석 표면에 선명하게 음각된 '수 리 영' 세 글자를 내려다보는 시선.\n\nLOCATION (lock): The marker-stone clearing deep in the forest, among several stones all bearing the same name.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Settle just behind 김지후(강수리영)'s shoulder at close range, using her shoulder and lowered head as a near-edge frame for the marker. The inscribed face occupies the central lower portion without exceeding a natural scale, and the three characters “수 리 영” sit precisely at the endpoint of her downward eyeline.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 비석의 ‘수 리 영’ 음각면 in the lower-center of the frame, midground; 김지후(강수리영) in the middle-left of the frame, foreground, looks toward inscribed marker face.\n- KEY BACKGROUND ELEMENTS: 낡은 비석 (old, with the name clearly incised) — Its inscribed front face is angled toward camera, clearly showing the recessed Korean characters ‘수 리 영’; used as Primary focal surface at the end of 김지후(강수리영)'s gaze; 주변의 다른 비석들 (present, each bearing the same name) — Only partial side and front planes are visible farther into the forest; used as Provides limited depth around the marker and shoulder framing.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime forest light renders the inscription clearly while preserving ominous high contrast and subdued earthy, aged-ivory, and blue-black tones.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S46sh2__bgfirst_bg.png",
  "asset_id": "4da52381-2727-4330-b832-9c87c065d469",
  "input_asset_ids": [
   "2faea7d8-dc3d-4ff6-b620-a6a0475cb9a4",
   "58a4c873-0b94-49b4-ad83-7174209aff1b"
  ]
 },
 "S46sh2": {
  "input_fingerprint": "fc04731756326cec",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 낡은 비석 표면에 선명하게 음각된 '수 리 영' 세 글자를 내려다보는 시선.\n\nLOCATION (lock): The marker-stone clearing deep in the forest, among several stones all bearing the same name. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Settle just behind 김지후(강수리영)'s shoulder at close range, using her shoulder and lowered head as a near-edge frame for the marker. The inscribed face occupies the central lower portion without exceeding a natural scale, and the three characters “수 리 영” sit precisely at the endpoint of her downward eyeline.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 비석의 ‘수 리 영’ 음각면 in the lower-center of the frame, midground; 김지후(강수리영) in the middle-left of the frame, foreground, looks toward inscribed marker face.\n- KEY BACKGROUND ELEMENTS: 낡은 비석 (old, with the name clearly incised) — Its inscribed front face is angled toward camera, clearly showing the recessed Korean characters ‘수 리 영’; used as Primary focal surface at the end of 김지후(강수리영)'s gaze; 주변의 다른 비석들 (present, each bearing the same name) — Only partial side and front planes are visible farther into the forest; used as Provides limited depth around the marker and shoulder framing.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime forest light renders the inscription clearly while preserving ominous high contrast and subdued earthy, aged-ivory, and blue-black tones.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The forest stones continue to bear the name “Su Ri-young,” with rotting meat before them. Jihu’s necklace remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 낡은 비석 표면에 선명하게 음각된 '수 리 영' 세 글자를 내려다보는 시선.\n\nLOCATION (lock): The marker-stone clearing deep in the forest, among several stones all bearing the same name. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Settle just behind 김지후(강수리영)'s shoulder at close range, using her shoulder and lowered head as a near-edge frame for the marker. The inscribed face occupies the central lower portion without exceeding a natural scale, and the three characters “수 리 영” sit precisely at the endpoint of her downward eyeline.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 비석의 ‘수 리 영’ 음각면 in the lower-center of the frame, midground; 김지후(강수리영) in the middle-left of the frame, foreground, looks toward inscribed marker face.\n- KEY BACKGROUND ELEMENTS: 낡은 비석 (old, with the name clearly incised) — Its inscribed front face is angled toward camera, clearly showing the recessed Korean characters ‘수 리 영’; used as Primary focal surface at the end of 김지후(강수리영)'s gaze; 주변의 다른 비석들 (present, each bearing the same name) — Only partial side and front planes are visible farther into the forest; used as Provides limited depth around the marker and shoulder framing.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime forest light renders the inscription clearly while preserving ominous high contrast and subdued earthy, aged-ivory, and blue-black tones.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The forest stones continue to bear the name “Su Ri-young,” with rotting meat before them. Jihu’s necklace remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)가 낡은 비석 표면에 선명하게 음각된 '수 리 영' 세 글자를 내려다보는 시선.\n\nLOCATION (lock): The marker-stone clearing deep in the forest, among several stones all bearing the same name. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Settle just behind 김지후(강수리영)'s shoulder at close range, using her shoulder and lowered head as a near-edge frame for the marker. The inscribed face occupies the central lower portion without exceeding a natural scale, and the three characters “수 리 영” sit precisely at the endpoint of her downward eyeline.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 비석의 ‘수 리 영’ 음각면 in the lower-center of the frame, midground; 김지후(강수리영) in the middle-left of the frame, foreground, looks toward inscribed marker face.\n- KEY BACKGROUND ELEMENTS: 낡은 비석 (old, with the name clearly incised) — Its inscribed front face is angled toward camera, clearly showing the recessed Korean characters ‘수 리 영’; used as Primary focal surface at the end of 김지후(강수리영)'s gaze; 주변의 다른 비석들 (present, each bearing the same name) — Only partial side and front planes are visible farther into the forest; used as Provides limited depth around the marker and shoulder framing.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime forest light renders the inscription clearly while preserving ominous high contrast and subdued earthy, aged-ivory, and blue-black tones.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The forest stones continue to bear the name “Su Ri-young,” with rotting meat before them. Jihu’s necklace remains missing.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S46sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S46sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:813104>"
    },
    {
     "label": "PROP REFERENCE — 이끼 낀 ‘수리영’ 비석: the exact object appearing in this shot; match its look, material and wear exactly.",
     "path": "<bytes:1345414>"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_path_network_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:813104>"
    },
    {
     "label": "PROP REFERENCE — 이끼 낀 ‘수리영’ 비석: the exact object appearing in this shot; match its look, material and wear exactly.",
     "path": "<bytes:1345414>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "여성이 화면 좌측 전경에 뒤통수/측면으로 서 있고 머리는 아래로 숙여져 있으나, 시선의 종착점이 비석의 '수 리 영' 음각면이 아니라 프레임 좌하단(비석보다 훨씬 아래·앞쪽)을 향한다. 비석은 그녀의 오른쪽 중앙에 있고 얼굴이 향한 방향과 각도가 어긋나 시선 끝이 글자에 닿지 않는다. 배경의 다른 비석 3기도 같은 이름을 새기고 있어 방향성 자체는 로케이션과 부합.",
      "built_space": "숲속 비석 군집. 중앙 비석은 로케이션 사진의 지붕형 갓돌+기단 구조를 이끼 낀 석재 몸돌과 결합해 재현. 배경 비석 3기는 로케이션 사진의 검은 화강암 형태를 따르고 각각 '수리영' 각자. 기단·받침·지면 관계는 물리적으로 성립. 다만 정면 각도가 캐릭터의 시선축과 정렬되지 않음.",
      "entities": "인물: 긴 흑발의 젊은 한국 여성, 카키/브라운 재킷 — 캐릭터 레퍼런스와 부합(목걸이 없음, 캐리드 스테이트 준수). 비석: 이끼 낀 석재에 '수 리 영' 음각, 프롭 레퍼런스와 유사하나 프롭 레퍼런스의 소박한 무갓돌 형태 대신 갓돌이 얹혀 로케이션 사진 쪽에 가깝게 혼합. 썩은 고기는 보이지 않음.",
      "hard_violations": [],
      "physics": "인물은 지면에 서 있고 어깨·머리 무게가 자연스럽게 실림. 비석들은 기단 위에 안정적으로 놓여 있고 부유하는 물체 없음. 이끼·흙·솔잎 등 재질감은 실제 표면으로 읽힘."
     },
     {
      "label": "B",
      "direction": "여성이 좌측 전경에서 어깨 너머 구도로 고개를 숙이고 있으며, 얼굴 방향과 시선축이 화면 중앙 하부의 이끼 낀 비석 '수리영' 음각면에 정확히 떨어진다. 좌우 배경 비석에도 같은 이름이 각자되어 있어 시선 종착점과 주변 깊이가 지시대로 성립.",
      "built_space": "숲속 공터, 소나무 숲과 흙길이 로케이션 사진의 구조와 재질을 따름. 중앙 비석은 프롭 레퍼런스의 무갓돌·이끼 낀 거친 석재와 거의 동일. 좌우 비석은 부분 측·정면만 보이며 더 깊은 숲 쪽에 배치 — 프레임 레이아웃 지시와 일치. 스케일도 실제 크기 범위 내.",
      "entities": "인물: 긴 흑발을 뒤로 넘긴 20대 초반 한국 여성, 브라운 워크재킷 — 캐릭터 레퍼런스와 부합, 목걸이 없음(캐리드 스테이트 준수). 비석: 프롭 레퍼런스의 이끼·균열·아이보리 석재와 '수 리 영' 음각이 표면 굴곡을 따라 실제로 파여 보임. 썩은 고기는 프레임 밖.",
      "hard_violations": [],
      "physics": "인물은 몸을 앞으로 기울여 지면에 발을 딛고 선 자세로 무게가 성립. 비석은 흙과 기단에 박혀 있고 부유물 없음. 이끼와 석재의 광반응, 재킷의 주름 모두 물리적으로 읽힘."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "로케이션·인물 일치는 좋으나 숙인 머리의 시선 종착점이 비석 음각면이 아닌 좌하단 허공으로 빗나가고, 비석이 어깨 너머 프레이밍보다 우측으로 분리되어 '시선 끝에 수 리 영' 지시를 충족하지 못함."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "어깨 너머 근접 클로즈업, 하단 중앙의 음각면, 숙인 시선의 정확한 종착점, 프롭·로케이션 재질 일치까지 카메라·프레임 지시를 가장 충실히 구현."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "로케이션·인물 일치는 좋으나 숙인 머리의 시선 종착점이 비석 음각면이 아닌 좌하단 허공으로 빗나가고, 비석이 어깨 너머 프레이밍보다 우측으로 분리되어 '시선 끝에 수 리 영' 지시를 충족하지 못함."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "어깨 너머 근접 클로즈업, 하단 중앙의 음각면, 숙인 시선의 정확한 종착점, 프롭·로케이션 재질 일치까지 카메라·프레임 지시를 가장 충실히 구현."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "여성이 화면 좌측 전경에 뒤통수/측면으로 서 있고 머리는 아래로 숙여져 있으나, 시선의 종착점이 비석의 '수 리 영' 음각면이 아니라 프레임 좌하단(비석보다 훨씬 아래·앞쪽)을 향한다. 비석은 그녀의 오른쪽 중앙에 있고 얼굴이 향한 방향과 각도가 어긋나 시선 끝이 글자에 닿지 않는다. 배경의 다른 비석 3기도 같은 이름을 새기고 있어 방향성 자체는 로케이션과 부합.",
      "built_space": "숲속 비석 군집. 중앙 비석은 로케이션 사진의 지붕형 갓돌+기단 구조를 이끼 낀 석재 몸돌과 결합해 재현. 배경 비석 3기는 로케이션 사진의 검은 화강암 형태를 따르고 각각 '수리영' 각자. 기단·받침·지면 관계는 물리적으로 성립. 다만 정면 각도가 캐릭터의 시선축과 정렬되지 않음.",
      "entities": "인물: 긴 흑발의 젊은 한국 여성, 카키/브라운 재킷 — 캐릭터 레퍼런스와 부합(목걸이 없음, 캐리드 스테이트 준수). 비석: 이끼 낀 석재에 '수 리 영' 음각, 프롭 레퍼런스와 유사하나 프롭 레퍼런스의 소박한 무갓돌 형태 대신 갓돌이 얹혀 로케이션 사진 쪽에 가깝게 혼합. 썩은 고기는 보이지 않음.",
      "hard_violations": [],
      "physics": "인물은 지면에 서 있고 어깨·머리 무게가 자연스럽게 실림. 비석들은 기단 위에 안정적으로 놓여 있고 부유하는 물체 없음. 이끼·흙·솔잎 등 재질감은 실제 표면으로 읽힘."
     },
     {
      "label": "B",
      "direction": "여성이 좌측 전경에서 어깨 너머 구도로 고개를 숙이고 있으며, 얼굴 방향과 시선축이 화면 중앙 하부의 이끼 낀 비석 '수리영' 음각면에 정확히 떨어진다. 좌우 배경 비석에도 같은 이름이 각자되어 있어 시선 종착점과 주변 깊이가 지시대로 성립.",
      "built_space": "숲속 공터, 소나무 숲과 흙길이 로케이션 사진의 구조와 재질을 따름. 중앙 비석은 프롭 레퍼런스의 무갓돌·이끼 낀 거친 석재와 거의 동일. 좌우 비석은 부분 측·정면만 보이며 더 깊은 숲 쪽에 배치 — 프레임 레이아웃 지시와 일치. 스케일도 실제 크기 범위 내.",
      "entities": "인물: 긴 흑발을 뒤로 넘긴 20대 초반 한국 여성, 브라운 워크재킷 — 캐릭터 레퍼런스와 부합, 목걸이 없음(캐리드 스테이트 준수). 비석: 프롭 레퍼런스의 이끼·균열·아이보리 석재와 '수 리 영' 음각이 표면 굴곡을 따라 실제로 파여 보임. 썩은 고기는 프레임 밖.",
      "hard_violations": [],
      "physics": "인물은 몸을 앞으로 기울여 지면에 발을 딛고 선 자세로 무게가 성립. 비석은 흙과 기단에 박혀 있고 부유물 없음. 이끼와 석재의 광반응, 재킷의 주름 모두 물리적으로 읽힘."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "여성의 머리가 오른쪽 아래로 숙여져 있고, 시선의 종점에 프레임 중앙 하단의 이끼 낀 비석 '수 리 영' 음각면이 놓인다. 얼굴은 4분의 3 뒤통수/측면으로 보이며 눈꺼풀이 내려간 방향이 비석 상부-글자면과 일치한다. 배경 좌·우의 다른 비석들도 같은 이름을 새기고 있어 시선 종점이 명확하다.",
      "built_space": "숲속 비석 군락. 중앙 주 비석 1기, 좌측 원경 1기, 우측 1기(기단석 포함)로 구성되어 중복 구조물이나 불가능한 배치는 없다. 지면은 솔잎 깔린 오솔길이 오른쪽으로 이어져 로케이션 사진의 길 구조와 부합한다. 다만 주 비석은 프롭 레퍼런스(이끼 낀 거친 돌)를 따르고 로케이션 사진의 갓돌 얹은 검은 화강암 형태는 반영되지 않았다.",
      "entities": "인물은 20대 초반 한국 여성, 긴 검은 생머리, 캐릭터 레퍼런스의 어두운 브라운 워크재킷과 일치. 모자는 보이지 않으나 프레이밍상 문제 없음. 목걸이는 보이지 않아 'necklace remains missing' 조건과 부합. 비석은 이끼·균열이 있는 아이보리 석재로 프롭 레퍼런스와 재질·마모가 일치하며 '수 리 영' 세 글자가 표면 곡면을 따라 음각으로 파여 있다(스티커 느낌 없음).",
      "hard_violations": [],
      "physics": "여성은 서 있거나 살짝 몸을 기울인 상태로 어깨·상체가 프레임 좌측 근경에 실려 있으며, 프레임 밖 하단에서 몸통이 이어져 공중에 뜬 요소가 없다. 비석들은 지면/기단석 위에 안정적으로 서 있고 그림자와 접지가 자연스럽다."
     },
     {
      "label": "B",
      "direction": "여성의 머리는 아래로 숙여져 있으나 얼굴이 거의 정면 아래(프레임 하단 왼쪽)를 향하고, 비석은 화면 중앙에서 다소 오른쪽·상방에 떨어져 있어 시선 종점이 '수 리 영' 음각면과 정확히 만나지 않는다. 눈코가 머리카락에 가려져 시선 방향 자체가 불명확하다.",
      "built_space": "주 비석은 갓돌+이끼 낀 몸체+기단석으로 로케이션 사진의 구조를 반영. 배경 우측에 갓돌 얹은 검은 비석 3기가 나란히 서 있고 각각 '수리영'이 새겨져 있으나, 원경 비석의 글자가 거리 대비 지나치게 또렷하고 크기 감이 다소 균일해 배치가 세트처럼 읽힌다. 중복 구조물이나 불가능한 배치는 아니다.",
      "entities": "인물은 긴 검은 생머리에 어두운 재킷으로 캐릭터 레퍼런스와 대체로 일치하나 얼굴이 거의 보이지 않아 동일성 확인이 어렵다. 목걸이 없음(조건 부합). 주 비석의 몸체는 프롭 레퍼런스의 이끼 낀 돌과 유사하나, 위에 얹힌 갓돌 비례가 어색해 프롭과 로케이션의 혼합이 부자연스럽다. 글자는 음각으로 표면에 앉아 있다.",
      "hard_violations": [],
      "physics": "여성은 좌측 근경에 서 있으며 어깨선이 프레임 밖으로 이어져 지지에 문제 없음. 비석들은 기단석 위에 접지되어 있고 부유 요소 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "숙인 머리 어깨너머 근접 구도에서 시선 종점이 하단 중앙의 '수 리 영' 음각면과 정확히 일치하고, 프롭 레퍼런스의 이끼 낀 석재 질감과 클로즈업 스케일을 충실히 구현했다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "비석이 중경 오른쪽으로 밀려 시선 종점과 어긋나고 인물이 화면을 크게 차지해 '어깨 너머 근접 클로즈업' 지시가 약화됐으며, 원경 비석 글자가 거리 대비 과도하게 선명하다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "숙인 머리 어깨너머 근접 구도에서 시선 종점이 하단 중앙의 '수 리 영' 음각면과 정확히 일치하고, 프롭 레퍼런스의 이끼 낀 석재 질감과 클로즈업 스케일을 충실히 구현했다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "비석이 중경 오른쪽으로 밀려 시선 종점과 어긋나고 인물이 화면을 크게 차지해 '어깨 너머 근접 클로즈업' 지시가 약화됐으며, 원경 비석 글자가 거리 대비 과도하게 선명하다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "여성의 머리가 오른쪽 아래로 숙여져 있고, 시선의 종점에 프레임 중앙 하단의 이끼 낀 비석 '수 리 영' 음각면이 놓인다. 얼굴은 4분의 3 뒤통수/측면으로 보이며 눈꺼풀이 내려간 방향이 비석 상부-글자면과 일치한다. 배경 좌·우의 다른 비석들도 같은 이름을 새기고 있어 시선 종점이 명확하다.",
      "built_space": "숲속 비석 군락. 중앙 주 비석 1기, 좌측 원경 1기, 우측 1기(기단석 포함)로 구성되어 중복 구조물이나 불가능한 배치는 없다. 지면은 솔잎 깔린 오솔길이 오른쪽으로 이어져 로케이션 사진의 길 구조와 부합한다. 다만 주 비석은 프롭 레퍼런스(이끼 낀 거친 돌)를 따르고 로케이션 사진의 갓돌 얹은 검은 화강암 형태는 반영되지 않았다.",
      "entities": "인물은 20대 초반 한국 여성, 긴 검은 생머리, 캐릭터 레퍼런스의 어두운 브라운 워크재킷과 일치. 모자는 보이지 않으나 프레이밍상 문제 없음. 목걸이는 보이지 않아 'necklace remains missing' 조건과 부합. 비석은 이끼·균열이 있는 아이보리 석재로 프롭 레퍼런스와 재질·마모가 일치하며 '수 리 영' 세 글자가 표면 곡면을 따라 음각으로 파여 있다(스티커 느낌 없음).",
      "hard_violations": [],
      "physics": "여성은 서 있거나 살짝 몸을 기울인 상태로 어깨·상체가 프레임 좌측 근경에 실려 있으며, 프레임 밖 하단에서 몸통이 이어져 공중에 뜬 요소가 없다. 비석들은 지면/기단석 위에 안정적으로 서 있고 그림자와 접지가 자연스럽다."
     },
     {
      "label": "A",
      "direction": "여성의 머리는 아래로 숙여져 있으나 얼굴이 거의 정면 아래(프레임 하단 왼쪽)를 향하고, 비석은 화면 중앙에서 다소 오른쪽·상방에 떨어져 있어 시선 종점이 '수 리 영' 음각면과 정확히 만나지 않는다. 눈코가 머리카락에 가려져 시선 방향 자체가 불명확하다.",
      "built_space": "주 비석은 갓돌+이끼 낀 몸체+기단석으로 로케이션 사진의 구조를 반영. 배경 우측에 갓돌 얹은 검은 비석 3기가 나란히 서 있고 각각 '수리영'이 새겨져 있으나, 원경 비석의 글자가 거리 대비 지나치게 또렷하고 크기 감이 다소 균일해 배치가 세트처럼 읽힌다. 중복 구조물이나 불가능한 배치는 아니다.",
      "entities": "인물은 긴 검은 생머리에 어두운 재킷으로 캐릭터 레퍼런스와 대체로 일치하나 얼굴이 거의 보이지 않아 동일성 확인이 어렵다. 목걸이 없음(조건 부합). 주 비석의 몸체는 프롭 레퍼런스의 이끼 낀 돌과 유사하나, 위에 얹힌 갓돌 비례가 어색해 프롭과 로케이션의 혼합이 부자연스럽다. 글자는 음각으로 표면에 앉아 있다.",
      "hard_violations": [],
      "physics": "여성은 좌측 근경에 서 있으며 어깨선이 프레임 밖으로 이어져 지지에 문제 없음. 비석들은 기단석 위에 접지되어 있고 부유 요소 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 17
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "readings": [
   {
    "label": "A",
    "direction": "여성이 화면 좌측 전경에 뒤통수/측면으로 서 있고 머리는 아래로 숙여져 있으나, 시선의 종착점이 비석의 '수 리 영' 음각면이 아니라 프레임 좌하단(비석보다 훨씬 아래·앞쪽)을 향한다. 비석은 그녀의 오른쪽 중앙에 있고 얼굴이 향한 방향과 각도가 어긋나 시선 끝이 글자에 닿지 않는다. 배경의 다른 비석 3기도 같은 이름을 새기고 있어 방향성 자체는 로케이션과 부합.",
    "built_space": "숲속 비석 군집. 중앙 비석은 로케이션 사진의 지붕형 갓돌+기단 구조를 이끼 낀 석재 몸돌과 결합해 재현. 배경 비석 3기는 로케이션 사진의 검은 화강암 형태를 따르고 각각 '수리영' 각자. 기단·받침·지면 관계는 물리적으로 성립. 다만 정면 각도가 캐릭터의 시선축과 정렬되지 않음.",
    "entities": "인물: 긴 흑발의 젊은 한국 여성, 카키/브라운 재킷 — 캐릭터 레퍼런스와 부합(목걸이 없음, 캐리드 스테이트 준수). 비석: 이끼 낀 석재에 '수 리 영' 음각, 프롭 레퍼런스와 유사하나 프롭 레퍼런스의 소박한 무갓돌 형태 대신 갓돌이 얹혀 로케이션 사진 쪽에 가깝게 혼합. 썩은 고기는 보이지 않음.",
    "hard_violations": [],
    "physics": "인물은 지면에 서 있고 어깨·머리 무게가 자연스럽게 실림. 비석들은 기단 위에 안정적으로 놓여 있고 부유하는 물체 없음. 이끼·흙·솔잎 등 재질감은 실제 표면으로 읽힘."
   },
   {
    "label": "B",
    "direction": "여성이 좌측 전경에서 어깨 너머 구도로 고개를 숙이고 있으며, 얼굴 방향과 시선축이 화면 중앙 하부의 이끼 낀 비석 '수리영' 음각면에 정확히 떨어진다. 좌우 배경 비석에도 같은 이름이 각자되어 있어 시선 종착점과 주변 깊이가 지시대로 성립.",
    "built_space": "숲속 공터, 소나무 숲과 흙길이 로케이션 사진의 구조와 재질을 따름. 중앙 비석은 프롭 레퍼런스의 무갓돌·이끼 낀 거친 석재와 거의 동일. 좌우 비석은 부분 측·정면만 보이며 더 깊은 숲 쪽에 배치 — 프레임 레이아웃 지시와 일치. 스케일도 실제 크기 범위 내.",
    "entities": "인물: 긴 흑발을 뒤로 넘긴 20대 초반 한국 여성, 브라운 워크재킷 — 캐릭터 레퍼런스와 부합, 목걸이 없음(캐리드 스테이트 준수). 비석: 프롭 레퍼런스의 이끼·균열·아이보리 석재와 '수 리 영' 음각이 표면 굴곡을 따라 실제로 파여 보임. 썩은 고기는 프레임 밖.",
    "hard_violations": [],
    "physics": "인물은 몸을 앞으로 기울여 지면에 발을 딛고 선 자세로 무게가 성립. 비석은 흙과 기단에 박혀 있고 부유물 없음. 이끼와 석재의 광반응, 재킷의 주름 모두 물리적으로 읽힘."
   }
  ],
  "totals": {
   "A": 11,
   "B": 17
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "로케이션·인물 일치는 좋으나 숙인 머리의 시선 종착점이 비석 음각면이 아닌 좌하단 허공으로 빗나가고, 비석이 어깨 너머 프레이밍보다 우측으로 분리되어 '시선 끝에 수 리 영' 지시를 충족하지 못함."
   },
   {
    "label": "B",
    "score": 9,
    "verdict_ko": "어깨 너머 근접 클로즈업, 하단 중앙의 음각면, 숙인 시선의 정확한 종착점, 프롭·로케이션 재질 일치까지 카메라·프레임 지시를 가장 충실히 구현."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_forest_path_network_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:813104>"
   },
   {
    "label": "PROP REFERENCE — 이끼 낀 ‘수리영’ 비석: the exact object appearing in this shot; match its look, material and wear exactly.",
    "path": "<bytes:1345414>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "캐릭터 기준 이미지에 있는 진회색 캡 모자를 지후가 쓰지 않았습니다.",
     "fix_en": "Add the dark grey baseball cap from the character reference onto Jihu's head. Preserve Jihu's position, her jacket, the main stone with its text, the background stones, the forest setting, the lighting, and the exact camera framing.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "왼쪽 배경의 비석에 '수리영'이 아닌 오타(쑤리영)가 새겨져 있습니다.",
     "fix_en": "Cover the carved face of the left background stone entirely with thick green moss, or turn the stone so its blank side faces the camera, ensuring no incorrect lettering is legible. Preserve Jihu's pose and clothing, the main stone and its '수 리 영' text, the right background stone, the forest background, the lighting, and the framing.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "지시문에 명시된 '썩은 고기(rotting meat)'가 화면에 보이는 우측 비석 앞 바닥에 없습니다.",
     "fix_en": "Place a lump of dark red, raw rotting meat on the dirt ground immediately in front of the base of the right background stone. Preserve Jihu's pose and clothing, the main stone and its text, the left background stone, the forest environment, the lighting, and the camera framing.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the dark grey baseball cap from the character reference onto Jihu's head. Preserve Jihu's position, her jacket, the main stone with its text, the background stones, the forest setting, the lighting, and the exact camera framing.\n- Cover the carved face of the left background stone entirely with thick green moss, or turn the stone so its blank side faces the camera, ensuring no incorrect lettering is legible. Preserve Jihu's pose and clothing, the main stone and its '수 리 영' text, the right background stone, the forest background, the lighting, and the framing.\n- Place a lump of dark red, raw rotting meat on the dirt ground immediately in front of the base of the right background stone. Preserve Jihu's pose and clothing, the main stone and its text, the left background stone, the forest environment, the lighting, and the camera framing.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "시선·음각·로케이션은 맞지만 캐릭터 레퍼런스의 회색 볼캡이 빠지고 헤어가 다르게 바뀌었으며, CARRIED STATE가 명시한 '비석 앞 썩은 고기'가 전혀 없다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 프레이밍·아이라인·로케이션을 유지하면서 레퍼런스의 볼캡과 CARRIED STATE의 부패육을 정확히 되살렸고, 그 외 인물·비석·빛은 거의 변경 없이 보존되어 수리가 잃은 것보다 얻은 것이 크다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.528,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.528,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "시선·음각·로케이션은 맞지만 캐릭터 레퍼런스의 회색 볼캡이 빠지고 헤어가 다르게 바뀌었으며, CARRIED STATE가 명시한 '비석 앞 썩은 고기'가 전혀 없다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 프레이밍·아이라인·로케이션을 유지하면서 레퍼런스의 볼캡과 CARRIED STATE의 부패육을 정확히 되살렸고, 그 외 인물·비석·빛은 거의 변경 없이 보존되어 수리가 잃은 것보다 얻은 것이 크다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "여성은 프레임 왼쪽 전경에서 등을 보이며 고개를 아래로 숙이고 있고, 시선의 종착점은 중앙 하단의 이끼 낀 비석 정면에 음각된 '수 리 영' 세 글자다. 머리 기울기와 얼굴 각도가 그 글자면을 정확히 향한다. 오른쪽 뒤로 또 다른 '수 리 영' 비석이 부분적으로 보이고, 그 앞 땅에 붉은 썩은 고기 덩어리가 놓여 있다.",
      "built_space": "숲속 비석 군락. 중앙 비석 1기(정면 음각면이 카메라 쪽으로 살짝 틀어짐), 좌측 뒤 1기, 우측 뒤 1기 — 중복이나 여분 구조물 없음. 낙엽 깔린 오솔길, 소나무 줄기, 관목이 레퍼런스 장소의 재질·색조와 일치. 우측 비석에는 기단 콘크리트가 보이며 스케일이 자연스럽다.",
      "entities": "인물: 한국인 20대 초반 여성, 검은 긴 머리(뒤로 흘러내림), 캐릭터 레퍼런스의 회색 볼캡과 짙은 갈색 워크재킷 착용 — 레퍼런스와 일치. 목걸이는 보이지 않음(계약대로 없음). 비석: 프롭 레퍼런스와 같은 이끼 낀 회백색 화강암, 한글 '수리영' 음각이 표면 곡률·이끼를 따라 파여 있음. 썩은 고기: 붉은 살덩이로 지면에 놓임 — CARRIED STATE 충족.",
      "hard_violations": [],
      "physics": "여성은 지면에 무릎/발을 딛고 상체를 숙인 자세로 프레임 좌측 하단에 안정적으로 자리하며, 공중에 뜬 요소 없음. 비석들은 지면·기단 위에 세워져 있고 그림자와 접지가 자연스럽다. 고기 덩어리는 흙바닥 위에 놓여 무게감 있게 눌려 있다."
     },
     {
      "label": "B",
      "direction": "여성은 같은 위치에서 고개를 숙여 중앙 비석의 '수 리 영' 음각면을 내려다본다. 시선 종착점은 정확하다. 좌측 뒤 비석에도 '수 리 영'이, 우측 뒤 비석에도 같은 글자가 보인다.",
      "built_space": "동일한 숲속 비석 군락. 비석 3기, 중복 없음. 오솔길과 나무 배치는 A와 사실상 같으나 좌측 뒤 비석이 더 크게/가깝게 렌더되고 그 위 이끼가 사라져 밝은 회백색 돌로 바뀌었다. 우측 비석 앞의 썩은 고기가 사라져 CARRIED STATE 항목(썩은 고기)이 누락됨.",
      "entities": "인물: 한국인 20대 여성이나 캐릭터 레퍼런스의 회색 볼캡이 제거되었고, 머리는 매끈하게 빗어 넘긴 청흑색 헤어로 바뀌어 질감이 인공적(윤기 과다, CG 같은 결). 재킷은 동일. 비석은 프롭 레퍼런스와 일치하는 이끼 낀 화강암이며 음각도 양호.",
      "hard_violations": [],
      "physics": "여성은 지면에 상체를 숙인 자세로 지지되며 부유 요소 없음. 비석들은 기단/지면에 접지. 다만 머리카락이 한 덩어리로 굳어 보여 물리적 결이 다소 떨어진다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "숄더 너머 클로즈업, 하단 중앙 음각면에 시선 종착, 캐릭터 레퍼런스의 볼캡·재킷 일치, 계약이 요구한 '썩은 고기'까지 유지 — 요구사항을 가장 많이 충족한다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "구도는 유사하나 계약이 요청하지 않은 변경(레퍼런스의 회색 캡 제거, 인공적인 청흑색 헤어, 좌측 비석 이끼 소실 및 확대)이 발생했고 CARRIED STATE의 썩은 고기가 사라져 손실이 이득보다 크다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "숄더 너머 클로즈업, 하단 중앙 음각면에 시선 종착, 캐릭터 레퍼런스의 볼캡·재킷 일치, 계약이 요구한 '썩은 고기'까지 유지 — 요구사항을 가장 많이 충족한다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "구도는 유사하나 계약이 요청하지 않은 변경(레퍼런스의 회색 캡 제거, 인공적인 청흑색 헤어, 좌측 비석 이끼 소실 및 확대)이 발생했고 CARRIED STATE의 썩은 고기가 사라져 손실이 이득보다 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "여성은 프레임 왼쪽 전경에서 등을 보이며 고개를 아래로 숙이고 있고, 시선의 종착점은 중앙 하단의 이끼 낀 비석 정면에 음각된 '수 리 영' 세 글자다. 머리 기울기와 얼굴 각도가 그 글자면을 정확히 향한다. 오른쪽 뒤로 또 다른 '수 리 영' 비석이 부분적으로 보이고, 그 앞 땅에 붉은 썩은 고기 덩어리가 놓여 있다.",
      "built_space": "숲속 비석 군락. 중앙 비석 1기(정면 음각면이 카메라 쪽으로 살짝 틀어짐), 좌측 뒤 1기, 우측 뒤 1기 — 중복이나 여분 구조물 없음. 낙엽 깔린 오솔길, 소나무 줄기, 관목이 레퍼런스 장소의 재질·색조와 일치. 우측 비석에는 기단 콘크리트가 보이며 스케일이 자연스럽다.",
      "entities": "인물: 한국인 20대 초반 여성, 검은 긴 머리(뒤로 흘러내림), 캐릭터 레퍼런스의 회색 볼캡과 짙은 갈색 워크재킷 착용 — 레퍼런스와 일치. 목걸이는 보이지 않음(계약대로 없음). 비석: 프롭 레퍼런스와 같은 이끼 낀 회백색 화강암, 한글 '수리영' 음각이 표면 곡률·이끼를 따라 파여 있음. 썩은 고기: 붉은 살덩이로 지면에 놓임 — CARRIED STATE 충족.",
      "hard_violations": [],
      "physics": "여성은 지면에 무릎/발을 딛고 상체를 숙인 자세로 프레임 좌측 하단에 안정적으로 자리하며, 공중에 뜬 요소 없음. 비석들은 지면·기단 위에 세워져 있고 그림자와 접지가 자연스럽다. 고기 덩어리는 흙바닥 위에 놓여 무게감 있게 눌려 있다."
     },
     {
      "label": "A",
      "direction": "여성은 같은 위치에서 고개를 숙여 중앙 비석의 '수 리 영' 음각면을 내려다본다. 시선 종착점은 정확하다. 좌측 뒤 비석에도 '수 리 영'이, 우측 뒤 비석에도 같은 글자가 보인다.",
      "built_space": "동일한 숲속 비석 군락. 비석 3기, 중복 없음. 오솔길과 나무 배치는 A와 사실상 같으나 좌측 뒤 비석이 더 크게/가깝게 렌더되고 그 위 이끼가 사라져 밝은 회백색 돌로 바뀌었다. 우측 비석 앞의 썩은 고기가 사라져 CARRIED STATE 항목(썩은 고기)이 누락됨.",
      "entities": "인물: 한국인 20대 여성이나 캐릭터 레퍼런스의 회색 볼캡이 제거되었고, 머리는 매끈하게 빗어 넘긴 청흑색 헤어로 바뀌어 질감이 인공적(윤기 과다, CG 같은 결). 재킷은 동일. 비석은 프롭 레퍼런스와 일치하는 이끼 낀 화강암이며 음각도 양호.",
      "hard_violations": [],
      "physics": "여성은 지면에 상체를 숙인 자세로 지지되며 부유 요소 없음. 비석들은 기단/지면에 접지. 다만 머리카락이 한 덩어리로 굳어 보여 물리적 결이 다소 떨어진다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S46sh2__bgfirst_bg.png",
   "bg_asset_id": "4da52381-2727-4330-b832-9c87c065d469",
   "bg_record_key": "S46sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S47sh8": {
  "input_fingerprint": "29b4d102c8800569",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 훼손된 시신 주변에서 쪼그려 앉아 현장 조사를 하던 조 형사와 신 형사가 난감한 표정으로 바닥을 내려다보는 모습.\n\nLOCATION (lock): The cordoned section at the far end of the mainland harbor pier around a mutilated body. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Dolly in from low waist height inside the restricted area and angle downward into a compact triangular arrangement formed by 조 형사, 신 형사, and the body between them. Keep both detectives fully crouched on opposite sides in a wide composition, their shared downward eyeline converging on the remains while the police line and spectators stay secondary beyond them.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 폴리스라인 (set up to block access) — The warning side faces the crowd and remains obliquely visible behind the detectives; used as Defines the restricted perimeter behind the investigating detectives; 몰려든 구경꾼들 (gathered and photographing the scene); used as Forms a busy but subordinate background beyond the restricted area; 구경꾼들의 휴대폰 (being used for photographs) — Several screen sides face toward their users or the body as photographs are taken; used as Visible in spectators' hands as evidence of their intrusive behavior.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime dock light reveals the investigative details plainly while subdued contrast and drained earthy tones preserve the tragic severity.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same mainland pier, daylight, gathered onlookers, police line, and waterfront textures. Exclude the seated injured man and barrel-fire group from the focal area; show the detectives examining the mutilated body.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): The corpse is sprawled directly on the ground with the head and one arm torn away. The remaining torso, attached arm, and legs rest motionless against the surface.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The corpse is sprawled directly on the ground with the head and one arm torn away. The remaining torso, attached arm, and legs rest motionless against the surface. Inwoo still has one arm in a cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 조 형사 (Korean man in his 40s, short dark hair, approachable mature face); 신 형사 (Korean man in his 30s, short dark hair, alert angular face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 6,
   "C": 4
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "낮은 허리 높이 하향 앵글의 와이드에서 두 형사가 마주 보고 완전히 쪼그려 앉아 시선이 중앙 형체로 수렴하고, 폴리스라인과 촬영 중인 구경꾼이 후경에 종속적으로 배치되어 촬영지시를 가장 충실히 구현함.  ★위반: [gpt] 프롬프트에 없는 대형 시신 덮개를 추가해 핵심 대상인 머리와 한쪽 팔이 없는 훼손 시신을 완전히 대체·은폐했다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "삼각 구도와 시선 수렴은 성립하나 폴리스라인 문자가 화면을 가로지르며 전경적 존재감으로 커지고 파란 타포린이 현장을 완전히 가려 '훼손된 시신 주변 현장 조사'의 정보가 약해짐.  ★위반: [gpt] 프롬프트에 없는 대형 파란 방수포를 추가해 훼손 시신을 완전히 은폐하고 핵심 개체의 명시된 형태를 확인할 수 없게 했다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "구경꾼들이 촬영보다 카메라 정면을 응시해 '몰려들어 사진 찍는' 지시가 약하고, 폴리스라인에 뭉개진 왜곡 문자열이 반복 노출되며 신 형사 인물 일치도도 낮음.  ★위반: [claude-opus] 폴리스라인 테이프에 의미 없는 왜곡 문자열(PGEE LINE GROSEN ZOIN 등)이 반복 렌더링 — 화면에 뭉개진 텍스트가 도드라짐 / [gpt] 프롬프트에 없는 대형 시신 덮개를 추가해 머리와 한쪽 팔이 없는 훼손 시신을 완전히 은폐했다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S47sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 조 형사: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:957998>"
   },
   {
    "label": "CHARACTER REFERENCE — 신 형사: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:862922>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "크리티컬(Critical): 프롬프트에 시신이 머리와 한쪽 팔이 없는 상태로 바닥에 널브러져 있어야 한다고 명시되어 있으나, 생성된 이미지에서는 시신이 회색 방수포로 완전히 덮여 있어 요구된 신체 훼손 상태를 전혀 확인할 수 없습니다.",
     "fix_en": "Remove the grey tarp covering the body on the ground between the detectives. In that exact spot, render the exposed, mutilated corpse sprawled directly on the concrete: show a torso, one attached arm, and legs resting motionless, with the head and the other arm completely missing. Preserve the two crouching detectives, their poses and clothing, the background crowd, the police tape, and the concrete pier exactly as they appear.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "메이저(Major): 이전 샷의 고정된 배경 요소를 그대로 유지해야 함에도 불구하고, 이전 샷에는 없었던 철제 난간 구조물이 화면 왼쪽 부두 가장자리에 새롭게 나타났습니다.",
     "fix_en": "",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the grey tarp covering the body on the ground between the detectives. In that exact spot, render the exposed, mutilated corpse sprawled directly on the concrete: show a torso, one attached arm, and legs resting motionless, with the head and the other arm completely missing. Preserve the two crouching detectives, their poses and clothing, the background crowd, the police tape, and the concrete pier exactly as they appear.\n- \n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 형사가 프레임 좌·우에 쪼그려 앉아 각각 시선을 아래 중앙의 천 덮인 형체로 내려 보내고 있어, 시선 삼각형이 바닥 중앙에서 수렴한다. 우측 배경 구경꾼들의 휴대폰은 화면 뒷면이 카메라 쪽, 화면은 사용자 눈 쪽을 향해 촬영 자세를 취한다. 폴리스라인 테이프는 우측 배경에서 비스듬히 걸쳐 제한구역 경계를 만든다.",
      "built_space": "부두 구조는 이전 스틸과 일치 — 좌측 바다·계선주·로프, 중앙 소실점으로 뻗는 콘크리트 부두, 우측 창문 달린 콘크리트 관리동과 파란 드럼·박스, 가로등·전신주 배열까지 동일. 좌측 난간이 추가로 보이나 부두 구조로 자연스럽다. 촬영 높이는 허리 낮은 위치에서 약간 내려보는 각도로 명세와 부합.",
      "entities": "좌측 인물은 40대 한국 남성, 짧은 흑발, 온화한 성숙한 얼굴 — 조 형사 프로필과 일치. 우측은 30대, 각진 얼굴의 한국 남성 — 신 형사와 일치(참조는 사극 의상이나 현대 파트이므로 얼굴 기준). 구경꾼은 배경에 종속적으로 배치되고 휴대폰을 든 손이 명확하다. 중앙 형체는 천으로 덮여 훼손 상태가 직접 보이지는 않는다(계약의 '머리·한쪽 팔이 뜯겨나간' 상태가 시각적으로 확인 불가).",
      "hard_violations": [],
      "physics": "두 형사 모두 발바닥과 발끝이 지면에 닿고 무릎이 접힌 완전한 쪼그린 자세, 한쪽 팔은 무릎 위에 얹혀 하중이 명확하다. 중앙 형체는 지면에 완전히 놓여 있고 천이 중력에 따라 늘어져 주름과 접힘을 형성한다. 공중에 뜬 요소 없음."
     },
     {
      "label": "B",
      "direction": "두 형사의 시선은 중앙 바닥의 인물 쪽으로 내려가며 수렴한다. 배경 구경꾼 휴대폰 방향은 A와 동일하게 자연스럽다. 다만 중앙 인물이 머리가 붙은 채 정면 하늘을 향해 누워 있어, 형사들이 내려다보는 대상이 '훼손된 시신'이 아니라 온전한 인물로 읽힌다.",
      "built_space": "부두·관리동·전신주·폴리스라인 배치는 A와 사실상 동일하게 유지. 카메라 높이와 하향 각도도 유사. 좌측 난간, 로프, 파란 드럼 모두 이전 스틸과 부합.",
      "entities": "좌·우 형사의 얼굴·의상은 A와 거의 동일해 조 형사/신 형사 프로필에 부합. 문제는 중앙 인물 — 계약은 '머리와 한쪽 팔이 뜯겨나간' 훼손된 시신을 못박았는데, 여기서는 비니 쓴 머리와 양팔·양다리가 모두 온전하며 얼굴까지 드러나 있다. 게다가 이 인물의 비니·작업복은 이전 스틸에서 제외하라고 지정한 드럼통 화롯가 남자와 유사해, 제외 지시를 어긴 인물 추가로 읽힌다.",
      "hard_violations": [
       "필수 캐논 포즈 위반: 머리와 한쪽 팔이 뜯겨나간 상태여야 할 형체가 사지·두부가 모두 온전한 인물로 렌더링됨",
       "이전 스틸에서 제외하라고 명시된 비니 착용 인물이 중앙 피사체로 등장 — 지시되지 않은 인물 도입"
      ],
      "physics": "중앙 인물은 지면에 등을 대고 완전히 누워 있고 팔·다리가 바닥에 닿아 중력에 순응한다. 두 형사도 발과 무릎으로 지지되어 있다. 부양 요소는 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "낮은 허리 높이 하향 앵글의 와이드, 좌우 대칭 쪼그림과 바닥으로 수렴하는 시선 삼각형, 배경의 폴리스라인·구경꾼·휴대폰까지 명세대로 구현했고, 훼손 형체를 천으로 덮어 캐논 포즈를 직접 위반하지 않으면서 장소 락도 유지했다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "구도·장소는 A와 거의 같지만, 필수 캐논인 '머리와 한쪽 팔이 뜯겨나간' 형체를 두부·사지가 온전한 인물로 바꾸고 이전 스틸에서 제외하라고 한 비니 차림 인물을 중앙에 세워, 고친 것보다 잃은 것이 훨씬 크다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "낮은 허리 높이 하향 앵글의 와이드, 좌우 대칭 쪼그림과 바닥으로 수렴하는 시선 삼각형, 배경의 폴리스라인·구경꾼·휴대폰까지 명세대로 구현했고, 훼손 형체를 천으로 덮어 캐논 포즈를 직접 위반하지 않으면서 장소 락도 유지했다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "구도·장소는 A와 거의 같지만, 필수 캐논인 '머리와 한쪽 팔이 뜯겨나간' 형체를 두부·사지가 온전한 인물로 바꾸고 이전 스틸에서 제외하라고 한 비니 차림 인물을 중앙에 세워, 고친 것보다 잃은 것이 훨씬 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 형사가 프레임 좌·우에 쪼그려 앉아 각각 시선을 아래 중앙의 천 덮인 형체로 내려 보내고 있어, 시선 삼각형이 바닥 중앙에서 수렴한다. 우측 배경 구경꾼들의 휴대폰은 화면 뒷면이 카메라 쪽, 화면은 사용자 눈 쪽을 향해 촬영 자세를 취한다. 폴리스라인 테이프는 우측 배경에서 비스듬히 걸쳐 제한구역 경계를 만든다.",
      "built_space": "부두 구조는 이전 스틸과 일치 — 좌측 바다·계선주·로프, 중앙 소실점으로 뻗는 콘크리트 부두, 우측 창문 달린 콘크리트 관리동과 파란 드럼·박스, 가로등·전신주 배열까지 동일. 좌측 난간이 추가로 보이나 부두 구조로 자연스럽다. 촬영 높이는 허리 낮은 위치에서 약간 내려보는 각도로 명세와 부합.",
      "entities": "좌측 인물은 40대 한국 남성, 짧은 흑발, 온화한 성숙한 얼굴 — 조 형사 프로필과 일치. 우측은 30대, 각진 얼굴의 한국 남성 — 신 형사와 일치(참조는 사극 의상이나 현대 파트이므로 얼굴 기준). 구경꾼은 배경에 종속적으로 배치되고 휴대폰을 든 손이 명확하다. 중앙 형체는 천으로 덮여 훼손 상태가 직접 보이지는 않는다(계약의 '머리·한쪽 팔이 뜯겨나간' 상태가 시각적으로 확인 불가).",
      "hard_violations": [],
      "physics": "두 형사 모두 발바닥과 발끝이 지면에 닿고 무릎이 접힌 완전한 쪼그린 자세, 한쪽 팔은 무릎 위에 얹혀 하중이 명확하다. 중앙 형체는 지면에 완전히 놓여 있고 천이 중력에 따라 늘어져 주름과 접힘을 형성한다. 공중에 뜬 요소 없음."
     },
     {
      "label": "B",
      "direction": "두 형사의 시선은 중앙 바닥의 인물 쪽으로 내려가며 수렴한다. 배경 구경꾼 휴대폰 방향은 A와 동일하게 자연스럽다. 다만 중앙 인물이 머리가 붙은 채 정면 하늘을 향해 누워 있어, 형사들이 내려다보는 대상이 '훼손된 시신'이 아니라 온전한 인물로 읽힌다.",
      "built_space": "부두·관리동·전신주·폴리스라인 배치는 A와 사실상 동일하게 유지. 카메라 높이와 하향 각도도 유사. 좌측 난간, 로프, 파란 드럼 모두 이전 스틸과 부합.",
      "entities": "좌·우 형사의 얼굴·의상은 A와 거의 동일해 조 형사/신 형사 프로필에 부합. 문제는 중앙 인물 — 계약은 '머리와 한쪽 팔이 뜯겨나간' 훼손된 시신을 못박았는데, 여기서는 비니 쓴 머리와 양팔·양다리가 모두 온전하며 얼굴까지 드러나 있다. 게다가 이 인물의 비니·작업복은 이전 스틸에서 제외하라고 지정한 드럼통 화롯가 남자와 유사해, 제외 지시를 어긴 인물 추가로 읽힌다.",
      "hard_violations": [
       "필수 캐논 포즈 위반: 머리와 한쪽 팔이 뜯겨나간 상태여야 할 형체가 사지·두부가 모두 온전한 인물로 렌더링됨",
       "이전 스틸에서 제외하라고 명시된 비니 착용 인물이 중앙 피사체로 등장 — 지시되지 않은 인물 도입"
      ],
      "physics": "중앙 인물은 지면에 등을 대고 완전히 누워 있고 팔·다리가 바닥에 닿아 중력에 순응한다. 두 형사도 발과 무릎으로 지지되어 있다. 부양 요소는 없음."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "두 형사 모두 시선이 아래 중앙의 시신 쪽으로 수렴한다. 왼쪽 40대 조 형사는 고개를 숙여 시신 상반신 쪽을, 오른쪽 30대 신 형사도 아래로 눈을 내리깔아 시신을 본다. 배경 구경꾼들의 휴대폰 화면은 사용자 자신 쪽을 향하고 카메라 후면이 현장을 향해 촬영 중 — 방향 정합. 폴리스라인은 오른쪽 뒤로 비스듬히 걸쳐 구경꾼과 제한구역을 나눈다.",
      "built_space": "실외 부두. 좌측 바다·계선주·로프, 우측 콘크리트 관리동, 가로등·전주 원근, 이전 스틸의 재질과 고정물이 유지된다. 폴리스라인 테이프와 스탠션 1세트, 구경꾼 무리 한 덩어리 — 중복 구조물 없음.",
      "entities": "조 형사: 40대 한국 남성, 짧은 검은머리, 원만한 얼굴 — 참조와 부합. 신 형사: 30대 한국 남성, 각진 얼굴 — 부합. 바닥의 시신: 남성 작업복 차림으로 누워 있으나 머리와 한쪽 팔이 온전히 붙어 있어 '머리와 한쪽 팔이 뜯겨 나간' 캐논 포즈와 불일치(다만 시신 자체는 존재하고 훼손 묘사가 생략된 수준). 구경꾼 휴대폰 다수 확인.",
      "hard_violations": [],
      "physics": "시신은 등을 바닥에 완전히 붙이고 팔·다리 모두 지면에 늘어져 중력에 순응한다. 들린 부위 없음. 두 형사는 발바닥과 무릎으로 하중을 받는 정상 쪼그림 자세, 한 손은 무릎에 얹혀 지지된다. 공중에 뜬 물체 없음."
     },
     {
      "label": "B",
      "direction": "두 형사의 시선은 아래 중앙의 천 덮개 쪽으로 수렴한다 — 눈길 방향 자체는 맞다. 구경꾼 휴대폰은 A와 동일하게 후면을 현장으로 향한다. 폴리스라인도 우측 뒤 비스듬히 유지.",
      "built_space": "A와 사실상 동일한 부두 세트, 고정물·가로등·관리동·계선주 모두 동일. 구조물 중복 없음.",
      "entities": "조 형사·신 형사 얼굴과 의상은 A와 동일하게 참조 부합. 그러나 '훼손된 시신'이 방수포/천으로 완전히 덮여 신체가 전혀 보이지 않는다 — 계약이 지정한 시신(머리·한 팔이 뜯긴 채 바닥에 널브러진 몸)이 프레임에서 사라지고 천 더미로 대체됨. 계약이 요구하지 않은 소품(덮개천) 추가.",
      "hard_violations": [],
      "physics": "천은 바닥의 무언가 위에 드리워져 있고 형태가 사람 형상이라기보다 둥근 둔덕이라 시신의 부피·자세를 읽을 수 없다. 형사 두 명의 쪼그림 지지는 정상. 공중 부양 요소는 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "시신이 실제로 바닥에 중력대로 널브러져 있고 두 형사의 하향 시선이 그 위로 수렴하는 삼각 구도, 폴리스라인·구경꾼·휴대폰 배경까지 계약대로 성립 — 머리/한쪽 팔 결손 묘사가 생략된 점만 감점."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "시신을 천으로 완전히 덮어 '훼손된 시신 주변 현장 조사'라는 샷 텍스트와 캐논 포즈(머리·한 팔이 뜯긴 채 바닥에 놓인 몸)를 화면에서 지워버렸고, 계약에 없는 덮개천 소품을 추가해 얻은 것보다 잃은 것이 크다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "시신이 실제로 바닥에 중력대로 널브러져 있고 두 형사의 하향 시선이 그 위로 수렴하는 삼각 구도, 폴리스라인·구경꾼·휴대폰 배경까지 계약대로 성립 — 머리/한쪽 팔 결손 묘사가 생략된 점만 감점."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "시신을 천으로 완전히 덮어 '훼손된 시신 주변 현장 조사'라는 샷 텍스트와 캐논 포즈(머리·한 팔이 뜯긴 채 바닥에 놓인 몸)를 화면에서 지워버렸고, 계약에 없는 덮개천 소품을 추가해 얻은 것보다 잃은 것이 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "두 형사 모두 시선이 아래 중앙의 시신 쪽으로 수렴한다. 왼쪽 40대 조 형사는 고개를 숙여 시신 상반신 쪽을, 오른쪽 30대 신 형사도 아래로 눈을 내리깔아 시신을 본다. 배경 구경꾼들의 휴대폰 화면은 사용자 자신 쪽을 향하고 카메라 후면이 현장을 향해 촬영 중 — 방향 정합. 폴리스라인은 오른쪽 뒤로 비스듬히 걸쳐 구경꾼과 제한구역을 나눈다.",
      "built_space": "실외 부두. 좌측 바다·계선주·로프, 우측 콘크리트 관리동, 가로등·전주 원근, 이전 스틸의 재질과 고정물이 유지된다. 폴리스라인 테이프와 스탠션 1세트, 구경꾼 무리 한 덩어리 — 중복 구조물 없음.",
      "entities": "조 형사: 40대 한국 남성, 짧은 검은머리, 원만한 얼굴 — 참조와 부합. 신 형사: 30대 한국 남성, 각진 얼굴 — 부합. 바닥의 시신: 남성 작업복 차림으로 누워 있으나 머리와 한쪽 팔이 온전히 붙어 있어 '머리와 한쪽 팔이 뜯겨 나간' 캐논 포즈와 불일치(다만 시신 자체는 존재하고 훼손 묘사가 생략된 수준). 구경꾼 휴대폰 다수 확인.",
      "hard_violations": [],
      "physics": "시신은 등을 바닥에 완전히 붙이고 팔·다리 모두 지면에 늘어져 중력에 순응한다. 들린 부위 없음. 두 형사는 발바닥과 무릎으로 하중을 받는 정상 쪼그림 자세, 한 손은 무릎에 얹혀 지지된다. 공중에 뜬 물체 없음."
     },
     {
      "label": "A",
      "direction": "두 형사의 시선은 아래 중앙의 천 덮개 쪽으로 수렴한다 — 눈길 방향 자체는 맞다. 구경꾼 휴대폰은 A와 동일하게 후면을 현장으로 향한다. 폴리스라인도 우측 뒤 비스듬히 유지.",
      "built_space": "A와 사실상 동일한 부두 세트, 고정물·가로등·관리동·계선주 모두 동일. 구조물 중복 없음.",
      "entities": "조 형사·신 형사 얼굴과 의상은 A와 동일하게 참조 부합. 그러나 '훼손된 시신'이 방수포/천으로 완전히 덮여 신체가 전혀 보이지 않는다 — 계약이 지정한 시신(머리·한 팔이 뜯긴 채 바닥에 널브러진 몸)이 프레임에서 사라지고 천 더미로 대체됨. 계약이 요구하지 않은 소품(덮개천) 추가.",
      "hard_violations": [],
      "physics": "천은 바닥의 무언가 위에 드리워져 있고 형태가 사람 형상이라기보다 둥근 둔덕이라 시신의 부피·자세를 읽을 수 없다. 형사 두 명의 쪼그림 지지는 정상. 공중 부양 요소는 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 10
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S47sh1"
  }
 },
 "S49sh1": {
  "input_fingerprint": "a7f9cac484d552f2",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 깜깜한 밤바다 멀리 불을 밝힌 야간 조업 어선들이 떠 있는 풍경.\n\nLOCATION (lock): An outdoor table at a mainland harbor sashimi restaurant, overlooking the dark sea and distant fishing-boat lights. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From seated eye height at the outer edge of the restaurant seating area, hold a static wide view straight across the dark sea toward the distant horizon. Keep the water occupying the lower field and the night sky above, while the illuminated fishing boats remain small, separated points across the far distance rather than oversized foreground subjects.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 밤바다 (Dark); used as Provides the broad negative space that isolates the distant fleet; 야간 조업 어선들 (Operating with their lights illuminated) — The boats are seen at varied oblique angles as small silhouettes far across the water; used as Forms the distant visual focus across the horizon; 먼 수평선 (Visible in the distance); used as Separates the dark water from the sky and distributes the boats laterally.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The boats' distant working lights punctuate an otherwise dark blue-black night seascape with sparse high contrast.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in its cast while he drinks with the detectives.\n\nNO PEOPLE IN THIS SHOT: the shot text shows only the place and its state — no living person appears in frame, and no face, and no figure standing, sitting or passing through. Do not invent someone to occupy the space.\n\nOne thing this rule does NOT forbid: the hand or arm of whoever is doing what the shot text describes. If the moment is an object being held, carried, operated, read, aimed, worn or handed over, the hand doing it belongs in frame, resting on or gripping that object exactly as its real use requires. An object photographed floating unsupported, with the hand that must be holding it erased, is a worse failure than the person you were avoiding — render the hand.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "카메라는 착석 눈높이에서 어두운 바다를 정면으로 가로질러 먼 수평선을 향한다. 시선축이 수평선에 정확히 놓이고, 불 밝힌 어선 광점들이 수평선을 따라 좌우로 분산되어 있다. 화면 하단은 나무 테이블과 난간, 그 위로 바다·하늘.",
    "built_space": "실외 테이블 좌석부 가장자리: 나무 테이블 상판, 목재 난간, 플라스틱 의자 2개(우측·우측 끝)가 화면 하단에 정상적으로 놓여 있다. 의자는 테이블을 향하고 있고 중복된 구조물이나 불가능한 반사는 없다.",
    "entities": "프롬프트가 지정한 요소만 존재 — 밤바다, 먼 수평선, 다수의 작은 어선 불빛, 횟집 야외 테이블(재떨이, 유리컵, 젓가락, 냅킨). 사람은 없고 얼굴·실루엣도 없다. 어선은 전경화되지 않고 작은 점 크기로 유지.",
    "hard_violations": [],
    "physics": "테이블 위 컵·재떨이·젓가락은 모두 상판에 놓여 접지 그림자가 있다. 공중에 뜬 물체 없음. 물결과 광점 반사는 자연스러운 야간 노출 반응."
   },
   {
    "label": "B",
    "direction": "카메라는 다소 높은 각도에서 좌측 아래 좌석부를 내려다보며 바다를 비스듬히 본다. 착석 눈높이 정면 응시가 아니라 좌석 배열을 옆에서 바라보는 사선 구도. 수평선은 정확히 있으나 어선 몇 척이 다소 크게 보이고 주 시선축이 정면이 아니다.",
    "built_space": "스틸 의자 다수와 나무 테이블 2~3개, 스테인리스 난간이 좌하단에 보인다. 배치 자체는 모순이 없으나 카메라가 좌석 '가장자리 착석 눈높이'가 아닌 뒤·위쪽 지점에 있어 프롬프트가 못 박은 뷰포인트와 다르다.",
    "entities": "밤바다·수평선·어선 불빛은 있으나, 화면 상단과 하단에 'PHOTOREALISTIC FILM STILL', 샷 텍스트 한국어 문구, 'LOCATION PHOTOGRAPH ... South Korea' 라벨과 하단 자막이 삽입되어 있다. 즉 문자·주석·패널 띠가 프레임에 존재.",
    "hard_violations": [
     "상단 라벨/제목 텍스트와 하단 자막 등 문자·주석 삽입 (프롬프트의 텍스트 금지 위반)",
     "상하 검은 띠에 캡션을 얹은 콜라주/패널 형식 레이아웃"
    ],
    "physics": "테이블·의자는 바닥에 접지되어 있고 부유 물체는 없음. 물리적 문제는 없으나 텍스트 오버레이가 평면 그래픽으로 얹혀 있다."
   },
   {
    "label": "C",
    "direction": "시선은 어두운 바다 건너 수평선을 향하며 어선 광점들이 가로로 분산되어 있다. 다만 좌우 전경에 정박한 어선 선체가 크게 들어와 시선을 분산시키고, 프레임 하단은 초점 흐린 테이블/수저통이 크게 차지한다.",
    "built_space": "우측 하단에 수저통, 냅킨통, 목재 테이블 상판, 좌하단에 의자 등받이가 보인다. 좌우 양쪽에 계류된 대형 어선이 전경 요소로 들어와 있어 '먼 수평선의 작은 점' 규정과 충돌하며, 프롬프트가 명시하지 않은 근접 선박이 추가되었다.",
    "entities": "밤바다·수평선·조업 어선 불빛은 있음. 사람은 없음. 그러나 전경 좌우의 정박 어선은 배경 요소를 전경 존재로 확대한 셈이고, 어선들이 A보다 크게 렌더되어 '작은 분리된 점' 지시에서 벗어난다.",
    "hard_violations": [],
    "physics": "수저통·냅킨통·테이블은 상판에 접지, 선박은 수면에 부유. 부유하는 미지지 물체 없음."
   }
  ],
  "totals": {
   "A": 9,
   "B": 2,
   "C": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 9,
    "verdict_ko": "착석 눈높이에서 밤바다를 정면으로 가로지르는 정적 와이드, 어선 불빛이 수평선에 작은 점으로 분산되고 사람 없이 횟집 야외 테이블만 하단에 놓여 프롬프트 구도·스케일·조명을 가장 충실히 구현했다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "화면 상단 제목·로케이션 라벨과 하단 한국어 자막 등 텍스트/패널 오버레이가 삽입되어 '텍스트·주석 금지'를 정면 위반했고, 카메라도 착석 눈높이 정면이 아닌 사선 부감이다."
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "밤바다와 어선 불빛은 무난하나, 좌우 전경에 계류 어선이 크게 들어오고 어선들이 점보다 크게 렌더되어 '작은 분리된 점·전경 확대 금지' 지시에서 벗어난다."
   }
  ],
  "refs": [],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S51sh4": {
  "input_fingerprint": "c373110dab906c25",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 분노와 답답함이 섞인 얼굴로 강민숙(홍은미)을 향해 핏대를 세운 채 소리치는 고산들의 클로즈업.\n\nLOCATION (lock): The dim cave chamber beside the chained prisoner's position under a low electric bulb. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At 고산들's eye level on the side facing 강민숙(홍은미), complete the dolly-in as a tight close-up at the crest of his shout, holding his strained three-quarter face across most of the frame without becoming an extreme close-up. A soft fragment of 강민숙(홍은미)'s shoulder remains at the lower opposite edge, fixing the downward direction of his accusation and preserving the exchange axis.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 동굴 내부 (Dark beyond the illuminated figures); used as Falls out of focus behind 고산들 so his anger is isolated without erasing the location; 쇠사슬 (Taut as she recoils) — A short section angles away from her body toward the floor outside the close focal plane; used as Appears only near 강민숙(홍은미)'s lower-edge silhouette as evidence of continued restraint.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: The low-watt bulb illumination cuts 고산들's face into severe high contrast while the cave background remains dark.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same dim cave, low bulb, rough rock, chains, and the captive woman's fixed position. Exclude the full-body standing tableau; frame the man's angry outburst toward her.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk remains chained on the cave floor while Sandel confronts her.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (Korean man in his mid-40s, closely cropped dark hair, cold angular face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 9,
   "B": 7,
   "C": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 9,
    "verdict_ko": "눈높이 3/4 클로즈업, 아래쪽 반대편 가장자리의 여성 어깨 조각과 초점 밖 쇠사슬·자물쇠, 어두운 동굴 배경과 낮은 전구까지 카메라·프레임 지시를 가장 정확히 충족하며 핏대 선 고함 연기도 일치."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "방향축과 인물 동일성은 양호하나 배경이 지시보다 밝고 디테일이 살아 있으며 쇠사슬 증거가 프레임에 없고 상대 인물이 '어깨 조각'보다 크게 들어와 프레이밍 충실도가 A보다 낮음."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "강민숙 자리에 짧은 머리의 남성으로 읽히는 인물이 서 있는 눈높이로 배치되어 인물·결박 상태가 어긋나고, 낮은 전구와 고대비 조명도 빠졌으며 의상도 레퍼런스와 불일치.  ★위반: [claude-opus] 샷 텍스트가 지정하지 않은 인물(짧은 머리의 남성으로 읽히는 전경 인물)이 강민숙 자리에 들어와 있음 — 인물 오지정/추가 / [gpt] 강민숙 대신 짧은 머리의 다른 인물로 보이는 몸을 대면 상대로 삽입했다. / [gpt] 바닥에 고정되어 있어야 할 포로를 고산들과 거의 같은 눈높이의 위치에 배치해 확립된 좌석·신체 위치를 위반했다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S51sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903959>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "프롬프트에 쇠사슬은 여자의 실루엣 하단 가장자리 근처에만 나타나고 바닥을 향해야 한다고 명시되었으나, 전경의 여자 어깨 위로 거대한 자물쇠와 쇠사슬이 잘못 배치되어 있습니다.",
     "fix_en": "Remove the large padlock and chain from the woman's shoulder in the foreground, replacing them with the plain out-of-focus brown fabric of her shirt, while strictly preserving the man, his expression and clothing, the lighting, and the cave background.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "고산들의 얼굴 생김새(특히 턱선과 코)가 제공된 캐릭터 레퍼런스의 인물과 일치하지 않습니다 (주요 오류).",
     "fix_en": "",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the large padlock and chain from the woman's shoulder in the foreground, replacing them with the plain out-of-focus brown fabric of her shirt, while strictly preserving the man, his expression and clothing, the lighting, and the cave background.\n- \n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "요구된 타이트한 클로즈업, 눈높이 3/4 얼굴, 하단 대각 고함 방향을 모두 충족하고 배경 요소로 명시된 쇠사슬(자물쇠 포함)이 초점 밖 하단에 정확히 남아 구속 증거를 보존한다.  ★위반: [gpt] 이전 장면과 계약에 없는 큰 자물쇠를 새로 만들어 민숙의 어깨 위에 배치했다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "인물·조명·장소는 동일하지만 프레임이 약간 넓어져 클로즈업 강도가 약해지고, 계약이 명시한 쇠사슬 파편이 제거되어 '구속 지속'의 시각 증거를 잃은 손실이 더 크다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.571,
      "B": 1.75
     },
     "adjusted": {
      "A": 1.321,
      "B": 1.75
     },
     "violations": {
      "A": [
       "[gpt] 이전 장면과 계약에 없는 큰 자물쇠를 새로 만들어 민숙의 어깨 위에 배치했다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "B"
     },
     "agreed": false
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "요구된 타이트한 클로즈업, 눈높이 3/4 얼굴, 하단 대각 고함 방향을 모두 충족하고 배경 요소로 명시된 쇠사슬(자물쇠 포함)이 초점 밖 하단에 정확히 남아 구속 증거를 보존한다.  ★위반: [gpt] 이전 장면과 계약에 없는 큰 자물쇠를 새로 만들어 민숙의 어깨 위에 배치했다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "인물·조명·장소는 동일하지만 프레임이 약간 넓어져 클로즈업 강도가 약해지고, 계약이 명시한 쇠사슬 파편이 제거되어 '구속 지속'의 시각 증거를 잃은 손실이 더 크다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "얼굴 클로즈업 스케일·시선축·조명은 계약대로지만, KEY BACKGROUND ELEMENTS가 명시한 '강민숙 하단 실루엣 근처의 쇠사슬(구속 증거)'이 프레임에 전혀 없어 계약 항목 하나가 누락됐다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "A와 인물·의상·표정·배경·조명이 사실상 동일하게 유지된 채, 누락됐던 쇠사슬만 강민숙 하단 실루엣 근처 초점 밖 전경으로 정확히 추가되어 '계속된 구속' 증거를 충족 — 얻은 것은 계약 항목이고 잃은 것은 거의 없어 B가 우세.  ★위반: [gpt] 계약과 이전 장면에 없는 큰 검은 자물쇠를 여성의 어깨 앞에 새로 발명해 추가했다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.778,
      "B": 1.571
     },
     "adjusted": {
      "A": 1.778,
      "B": 1.321
     },
     "violations": {
      "B": [
       "[gpt] 계약과 이전 장면에 없는 큰 검은 자물쇠를 여성의 어깨 앞에 새로 발명해 추가했다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "A"
     },
     "agreed": false
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "얼굴 클로즈업 스케일·시선축·조명은 계약대로지만, KEY BACKGROUND ELEMENTS가 명시한 '강민숙 하단 실루엣 근처의 쇠사슬(구속 증거)'이 프레임에 전혀 없어 계약 항목 하나가 누락됐다."
     },
     {
      "label": "A",
      "score": 9,
      "verdict_ko": "A와 인물·의상·표정·배경·조명이 사실상 동일하게 유지된 채, 누락됐던 쇠사슬만 강민숙 하단 실루엣 근처 초점 밖 전경으로 정확히 추가되어 '계속된 구속' 증거를 충족 — 얻은 것은 계약 항목이고 잃은 것은 거의 없어 B가 우세.  ★위반: [gpt] 계약과 이전 장면에 없는 큰 검은 자물쇠를 여성의 어깨 앞에 새로 발명해 추가했다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 17,
     "B": 13
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S51sh2"
  }
 },
 "S52sh1": {
  "input_fingerprint": "1de8886bb0aeee8e",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 숲속 공터 한가운데 검붉은 피가 잔뜩 묻은 거대한 바위가 기괴하게 서 있는 전경.\n\nLOCATION (lock): A ritual clearing deep in the forest, dominated by a towering boulder smeared with dark animal blood. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From the clearing's edge, hold a static, standing-eye-level wide view directly along the central axis toward the immense rock. Keep its full height centered in the midground, with the open clearing surrounding it and the dark-red hand-smeared blood traces remaining legible across the rock face.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 거대한 바위 (성인 키 서너 배 높이로 서 있으며 곳곳에 동물의 피 흔적이 묻어 있음) — Its broad blood-marked face is presented head-on to the camera; used as Primary central landmark and visual focus of the establishing composition; 숲속 공터 (산속 숲 한가운데가 공터처럼 트여 있음); used as Provides negative space around the rock and establishes the isolated ritual site; 공터 주변의 숲 (공터를 둘러싸고 있음); used as Forms the dark outer boundary of the composition.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key night ambience renders the clearing in blue-black and earthy neutrals while preserving the blood as a restrained dark-red accent.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The giant ritual rock remains heavily marked with dark red animal blood.\n\nPEOPLE: the SHOT TEXT alone decides who is visible in this shot. People known to appear somewhere in this scene: 고산들 (Korean man in his mid-40s, closely cropped dark hair, cold angular face). That list is scene-level, not a cast list for this frame — it may name someone this shot does not show, and it may omit someone this shot does show. If the shot text names a person who is not on the list, draw that person exactly as the shot text describes them; the list does not override the shot text. Never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 9,
   "C": 7,
   "B": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 9,
    "verdict_ko": "공터 가장자리 정면 아이레벨 와이드, 바위 전체 높이 중앙 배치, 검붉은 피가 요철을 따라 젖은 액체로 안착 — 야간 저조도 톤과 스케일까지 지정에 가장 충실."
   },
   {
    "label": "C",
    "score": 7,
    "verdict_ko": "정면 축·중앙 배치·야간 톤은 맞으나 바위 표면이 매끈하고 피가 넓게 도포돼 '손으로 문지른 흔적'의 판독성과 재질 실감이 떨어진다."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "손자국 흔적은 가장 뚜렷하지만 바위 높이가 '성인 키 서너 배'에 못 미치고 화면 지배력이 약하며, 피가 마른 그래픽처럼 얹혀 물질감이 부족하다."
   }
  ],
  "refs": [],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "플레이트만 (배경 전용)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S53sh1::bgfirst_bg": {
  "input_fingerprint": "aacbf1f520c7dba4",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 밝은 대낮의 선착장에 정박한 낚시 어선 위에서 황봉업(황 씨)이 불만 가득한 표정으로 꼿꼿하게 선 모습.\n\nLOCATION (lock): The deck of a fishing boat moored at the mainland harbor pier in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From standing eye level on the pier, hold a static wide composition facing 황봉업(황 씨) across the short gap to the boat. Place him rigidly in the middle-right aboard the vessel, leaving the pier-to-boat separation visible on the left so his refusal reads as a territorial barrier.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 정박한 낚시 어선 (선착장에 정박해 있음) — The pier-facing side of the boat is visible, with 황봉업(황 씨) positioned aboard it; used as Elevates 황봉업(황 씨) above the pier and establishes the contested means of passage; 선착장 (낚시 어선과 맞닿아 있음) — Its length leads from the lower-left foreground toward the moored boat; used as Creates the foreground approach line toward the boat.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Bright daytime ambient light keeps the confrontation plainly exposed, tempered by restrained earthy and blue-neutral color.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 밝은 대낮의 선착장에 정박한 낚시 어선 위에서 황봉업(황 씨)이 불만 가득한 표정으로 꼿꼿하게 선 모습.\n\nLOCATION (lock): The deck of a fishing boat moored at the mainland harbor pier in daylight.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From standing eye level on the pier, hold a static wide composition facing 황봉업(황 씨) across the short gap to the boat. Place him rigidly in the middle-right aboard the vessel, leaving the pier-to-boat separation visible on the left so his refusal reads as a territorial barrier.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 정박한 낚시 어선 (선착장에 정박해 있음) — The pier-facing side of the boat is visible, with 황봉업(황 씨) positioned aboard it; used as Elevates 황봉업(황 씨) above the pier and establishes the contested means of passage; 선착장 (낚시 어선과 맞닿아 있음) — Its length leads from the lower-left foreground toward the moored boat; used as Creates the foreground approach line toward the boat.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Bright daytime ambient light keeps the confrontation plainly exposed, tempered by restrained earthy and blue-neutral color.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S53sh1__bgfirst_bg.png",
  "asset_id": "fec13fda-50d3-4945-9920-0a88df918669",
  "input_asset_ids": [
   "78c2b044-a7cf-4d6d-af85-ff3631e54ab9",
   "a211d2e1-5029-4f1e-afab-471db46f09bb"
  ]
 },
 "S53sh1": {
  "input_fingerprint": "cf290e3b3cb9074d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 대낮의 선착장에 정박한 낚시 어선 위에서 황봉업(황 씨)이 불만 가득한 표정으로 꼿꼿하게 선 모습.\n\nLOCATION (lock): The deck of a fishing boat moored at the mainland harbor pier in daylight. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From standing eye level on the pier, hold a static wide composition facing 황봉업(황 씨) across the short gap to the boat. Place him rigidly in the middle-right aboard the vessel, leaving the pier-to-boat separation visible on the left so his refusal reads as a territorial barrier.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 정박한 낚시 어선 (선착장에 정박해 있음) — The pier-facing side of the boat is visible, with 황봉업(황 씨) positioned aboard it; used as Elevates 황봉업(황 씨) above the pier and establishes the contested means of passage; 선착장 (낚시 어선과 맞닿아 있음) — Its length leads from the lower-left foreground toward the moored boat; used as Creates the foreground approach line toward the boat.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Bright daytime ambient light keeps the confrontation plainly exposed, tempered by restrained earthy and blue-neutral color.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast as he confronts Hwang at the dock.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 황봉업(황 씨) (Korean man in his mid-50s, dark graying hair, broad genial face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 대낮의 선착장에 정박한 낚시 어선 위에서 황봉업(황 씨)이 불만 가득한 표정으로 꼿꼿하게 선 모습.\n\nLOCATION (lock): The deck of a fishing boat moored at the mainland harbor pier in daylight. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From standing eye level on the pier, hold a static wide composition facing 황봉업(황 씨) across the short gap to the boat. Place him rigidly in the middle-right aboard the vessel, leaving the pier-to-boat separation visible on the left so his refusal reads as a territorial barrier.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 정박한 낚시 어선 (선착장에 정박해 있음) — The pier-facing side of the boat is visible, with 황봉업(황 씨) positioned aboard it; used as Elevates 황봉업(황 씨) above the pier and establishes the contested means of passage; 선착장 (낚시 어선과 맞닿아 있음) — Its length leads from the lower-left foreground toward the moored boat; used as Creates the foreground approach line toward the boat.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Bright daytime ambient light keeps the confrontation plainly exposed, tempered by restrained earthy and blue-neutral color.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast as he confronts Hwang at the dock.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 황봉업(황 씨) (Korean man in his mid-50s, dark graying hair, broad genial face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 밝은 대낮의 선착장에 정박한 낚시 어선 위에서 황봉업(황 씨)이 불만 가득한 표정으로 꼿꼿하게 선 모습.\n\nLOCATION (lock): The deck of a fishing boat moored at the mainland harbor pier in daylight. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From standing eye level on the pier, hold a static wide composition facing 황봉업(황 씨) across the short gap to the boat. Place him rigidly in the middle-right aboard the vessel, leaving the pier-to-boat separation visible on the left so his refusal reads as a territorial barrier.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 정박한 낚시 어선 (선착장에 정박해 있음) — The pier-facing side of the boat is visible, with 황봉업(황 씨) positioned aboard it; used as Elevates 황봉업(황 씨) above the pier and establishes the contested means of passage; 선착장 (낚시 어선과 맞닿아 있음) — Its length leads from the lower-left foreground toward the moored boat; used as Creates the foreground approach line toward the boat.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Bright daytime ambient light keeps the confrontation plainly exposed, tempered by restrained earthy and blue-neutral color.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains in a cast as he confronts Hwang at the dock.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 황봉업(황 씨) (Korean man in his mid-50s, dark graying hair, broad genial face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S53sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S53sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 황봉업(황 씨): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:950885>"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_mainland_harbor_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 황봉업(황 씨): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:950885>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 선착장 위 눈높이에서 배를 향하고 있으나, 프레임 좌측 전경에 팔걸이(슬링)를 한 젊은 남자가 뒷모습으로 서서 배 위 남자를 바라보고 있다. 배 위 남자(황 씨로 추정)는 조타실 옆에 서서 선착장 쪽(좌측 전방)을 바라본다. 시선 방향 자체는 서로 맞물리지만, 샷 텍스트는 황 씨 한 사람만을 지시한다.",
      "built_space": "낚시 어선의 조타실, 난간, 집어등 줄, 그물 릴 등이 배치되어 있고 선착장은 좌측에서 뒤로 이어진다. 레퍼런스의 콘크리트 방파제·계선주·타이어 펜더·원경 섬은 어느 정도 반영되었으나 붉은 등대가 매우 작게 좌측 끝에 흐릿하게 존재. 배는 선착장 우측에 정박, 접안면이 카메라를 향해 보인다 — 구조 자체는 성립.",
      "entities": "황 씨: 50대 중반 한국 남성으로 보이나 캐릭터 레퍼런스의 남색 니트+고무 앞치마+장화 복장이 아니라 회녹색 작업복 상하의를 입었고, 얼굴 유사도도 낮은 편(더 젊고 마른 인상). 추가로 캐스트/슬링을 한 인우 캐릭터가 실제 인물로 프레임에 등장 — 프롬프트는 '샷 텍스트가 보여주지 않는 인물은 절대 추가하지 말 것'이며 등장 가능 인물은 황봉업 한 명뿐이라고 못박음.",
      "hard_violations": [
       "샷 텍스트에 없는 인물(팔 슬링을 한 남성)을 전경에 추가 — PEOPLE 조항 '황봉업 외 누구도 안 됨' 위반",
       "황 씨가 화면 중앙-우측이 아니라 중앙 부근이며, 추가 인물이 좌측 여백(선착장-배 간극)을 점유해 지정된 구도 논리 훼손"
      ],
      "physics": "두 인물 모두 발이 각각 선착장 콘크리트와 갑판에 닿아 체중을 지지한다. 배는 물에 떠 있고 계류 로프와 펜더가 접안 상태를 지지. 부유하거나 지지 없는 물체는 없음."
     },
     {
      "label": "B",
      "direction": "카메라는 선착장 위 서 있는 눈높이에서 짧은 간격 너머 배를 정면으로 바라본다. 황 씨는 갑판 위에 서서 시선을 카메라/선착장 쪽(좌측 약간 아래)으로 곧게 향하고 있으며, 불만 섞인 굳은 표정. 시선 대상은 선착장에 선 상대(프레임 밖) 방향으로 정확히 떨어진다.",
      "built_space": "낚시 어선의 조타실, 안테나 마스트, 배기 연통, 그물더미, 릴이 있고 선체 접안면이 카메라를 향한다. 선착장은 좌측 하단 전경에서 뒤쪽 배 쪽으로 길게 뻗어 접근선을 만들고, 계선주·타이어와 노란 펜더·붉은 등대·원경 섬·바위 방파제 등 레퍼런스 고정 요소가 잘 일치. 조타실은 하나, 구조 중복 없음.",
      "entities": "황 씨: 한국인 50대 남성, 짙은 머리, 넓은 얼굴형으로 캐릭터 레퍼런스와 상당히 근접. 복장도 레퍼런스의 남색 니트 + 낡은 고무 앞치마 + 어두운 바지/장화를 그대로 재현. 프레임 내 다른 인물 없음 — 샷 텍스트 요구와 일치.",
      "hard_violations": [],
      "physics": "황 씨는 두 발로 갑판 위에 서 있고 무게가 다리에 실려 있다. 배는 계류 로프와 펜더로 부두에 접안, 물 위에 정상적으로 떠 있음. 공중에 지지 없이 뜬 물체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "샷 텍스트가 지시하지 않은 슬링 착용 남성을 전경에 추가해 'PEOPLE — 황봉업 외 절대 금지' 조항을 정면 위반했고, 황 씨 의상·얼굴도 캐릭터 레퍼런스와 어긋난다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "황 씨 단독으로 갑판 중앙-우측에 꼿꼿이 서서 선착장 쪽을 노려보고, 좌측에 부두-배 간극과 접근선을 남긴 와이드 구도·대낮 광선·레퍼런스 복장/장소 요소가 모두 부합한다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "샷 텍스트가 지시하지 않은 슬링 착용 남성을 전경에 추가해 'PEOPLE — 황봉업 외 절대 금지' 조항을 정면 위반했고, 황 씨 의상·얼굴도 캐릭터 레퍼런스와 어긋난다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "황 씨 단독으로 갑판 중앙-우측에 꼿꼿이 서서 선착장 쪽을 노려보고, 좌측에 부두-배 간극과 접근선을 남긴 와이드 구도·대낮 광선·레퍼런스 복장/장소 요소가 모두 부합한다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 선착장 위 눈높이에서 배를 향하고 있으나, 프레임 좌측 전경에 팔걸이(슬링)를 한 젊은 남자가 뒷모습으로 서서 배 위 남자를 바라보고 있다. 배 위 남자(황 씨로 추정)는 조타실 옆에 서서 선착장 쪽(좌측 전방)을 바라본다. 시선 방향 자체는 서로 맞물리지만, 샷 텍스트는 황 씨 한 사람만을 지시한다.",
      "built_space": "낚시 어선의 조타실, 난간, 집어등 줄, 그물 릴 등이 배치되어 있고 선착장은 좌측에서 뒤로 이어진다. 레퍼런스의 콘크리트 방파제·계선주·타이어 펜더·원경 섬은 어느 정도 반영되었으나 붉은 등대가 매우 작게 좌측 끝에 흐릿하게 존재. 배는 선착장 우측에 정박, 접안면이 카메라를 향해 보인다 — 구조 자체는 성립.",
      "entities": "황 씨: 50대 중반 한국 남성으로 보이나 캐릭터 레퍼런스의 남색 니트+고무 앞치마+장화 복장이 아니라 회녹색 작업복 상하의를 입었고, 얼굴 유사도도 낮은 편(더 젊고 마른 인상). 추가로 캐스트/슬링을 한 인우 캐릭터가 실제 인물로 프레임에 등장 — 프롬프트는 '샷 텍스트가 보여주지 않는 인물은 절대 추가하지 말 것'이며 등장 가능 인물은 황봉업 한 명뿐이라고 못박음.",
      "hard_violations": [
       "샷 텍스트에 없는 인물(팔 슬링을 한 남성)을 전경에 추가 — PEOPLE 조항 '황봉업 외 누구도 안 됨' 위반",
       "황 씨가 화면 중앙-우측이 아니라 중앙 부근이며, 추가 인물이 좌측 여백(선착장-배 간극)을 점유해 지정된 구도 논리 훼손"
      ],
      "physics": "두 인물 모두 발이 각각 선착장 콘크리트와 갑판에 닿아 체중을 지지한다. 배는 물에 떠 있고 계류 로프와 펜더가 접안 상태를 지지. 부유하거나 지지 없는 물체는 없음."
     },
     {
      "label": "B",
      "direction": "카메라는 선착장 위 서 있는 눈높이에서 짧은 간격 너머 배를 정면으로 바라본다. 황 씨는 갑판 위에 서서 시선을 카메라/선착장 쪽(좌측 약간 아래)으로 곧게 향하고 있으며, 불만 섞인 굳은 표정. 시선 대상은 선착장에 선 상대(프레임 밖) 방향으로 정확히 떨어진다.",
      "built_space": "낚시 어선의 조타실, 안테나 마스트, 배기 연통, 그물더미, 릴이 있고 선체 접안면이 카메라를 향한다. 선착장은 좌측 하단 전경에서 뒤쪽 배 쪽으로 길게 뻗어 접근선을 만들고, 계선주·타이어와 노란 펜더·붉은 등대·원경 섬·바위 방파제 등 레퍼런스 고정 요소가 잘 일치. 조타실은 하나, 구조 중복 없음.",
      "entities": "황 씨: 한국인 50대 남성, 짙은 머리, 넓은 얼굴형으로 캐릭터 레퍼런스와 상당히 근접. 복장도 레퍼런스의 남색 니트 + 낡은 고무 앞치마 + 어두운 바지/장화를 그대로 재현. 프레임 내 다른 인물 없음 — 샷 텍스트 요구와 일치.",
      "hard_violations": [],
      "physics": "황 씨는 두 발로 갑판 위에 서 있고 무게가 다리에 실려 있다. 배는 계류 로프와 펜더로 부두에 접안, 물 위에 정상적으로 떠 있음. 공중에 지지 없이 뜬 물체 없음."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "황봉업은 배 갑판 위에서 카메라(부두 쪽) 정면을 향해 서 있고 시선은 부두 눈높이 카메라 방향, 즉 대치 상대가 있을 왼쪽 부두 쪽을 향한다. 배의 뱃머리는 화면 왼쪽 아래(부두 쪽)를 향하고 부두-배 사이의 물 간격이 화면 중앙~왼쪽에 보인다. 인물은 화면 중앙 우측에 배치되어 지시와 일치.",
      "built_space": "낚시 어선 한 척: 조타실 1개, 안테나/레이더 마스트, 배기 연통, 그물 릴, 계류 로프와 노란 펜더가 부두 볼라드에 걸려 있다. 부두는 참조 사진과 동일한 콘크리트 방파제, 낮은 연석, 낡은 볼라드, 배경의 붉은 등대와 바위 방파제, 섬 능선이 일치. 중복된 구조물이나 여분의 조타실 없음.",
      "entities": "인물은 1명뿐 — 50대 한국 남성, 검은 머리, 넓은 얼굴, 남색 니트 위에 낡은 고무 앞치마로 캐릭터 레퍼런스와 복장·얼굴이 잘 일치. 표정은 웃음기 없이 굳어 불만스러운 인상. 프롬프트가 금지한 추가 인물 없음.",
      "hard_violations": [],
      "physics": "두 발이 갑판 위에 확실히 놓여 꼿꼿하게 직립, 팔은 몸 옆으로 자연스럽게 내려져 있음. 배는 물에 떠 있고 로프·펜더로 부두에 고정되어 지지 관계가 명확. 공중에 떠 있는 물체 없음."
     },
     {
      "label": "B",
      "direction": "왼쪽 부두 위 인물(팔걸이를 한 젊은 남성)은 배 쪽을 바라보고, 배 위 남성은 부두 쪽을 향해 서 있다. 카메라는 부두 눈높이 광각. 그러나 프롬프트가 명시한 '황봉업만' 프레임에 들어가야 하는 지시를 어기고 인우로 보이는 인물이 전경에 추가됨.",
      "built_space": "어선 1척: 조타실 1개, 집어등 줄, 그물 릴, 펜더, 계류 로프 — 구조 자체는 정합. 부두는 참조와 유사하나 붉은 등대가 매우 작게 좌측 멀리 배치. 배 위 인물은 조타실 옆 갑판에 서 있음.",
      "entities": "인물 2명: (1) 부두의 검은 재킷·팔 슬링 착용 청년 — 프롬프트가 '절대 추가하지 말라'고 한 인물. (2) 배 위 남성은 국방색 작업복 차림으로 캐릭터 레퍼런스의 남색 니트+고무 앞치마 복장과 불일치하고 얼굴 크기가 작아 동일 인물 확인이 어려움. 또한 화면 중앙 우측이 아니라 인물이 중앙보다 우측 조타실 옆이지만 크기가 매우 작음.",
      "hard_violations": [
       "프롬프트가 명시적으로 금지한 추가 인물(인우로 보이는 팔 깁스 남성)이 프레임 안에 등장 — invented/extra person"
      ],
      "physics": "두 인물 모두 두 발로 지면(부두/갑판)에 지지, 부양 문제 없음. 배는 로프와 펜더로 부두에 계류되어 있음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "황봉업 단독으로 배 갑판 중앙우측에 꼿꼿이 서서 부두 쪽을 노려보고, 부두-배 간격이 좌측에 보이며 등대·방파제 등 로케이션 구조도 참조와 일치 — 인물 스케일이 다소 크지만 와이드 구성과 캐릭터 복장·얼굴 일치가 우수."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "'황봉업 외 인물 추가 금지' 규정을 어기고 전경에 슬링을 한 인물을 넣은 하드 위반이며, 배 위 인물의 복장도 레퍼런스(남색 니트+고무 앞치마)와 어긋나고 인물이 너무 작아 '불만 가득한 표정'이 읽히지 않음."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "황봉업 단독으로 배 갑판 중앙우측에 꼿꼿이 서서 부두 쪽을 노려보고, 부두-배 간격이 좌측에 보이며 등대·방파제 등 로케이션 구조도 참조와 일치 — 인물 스케일이 다소 크지만 와이드 구성과 캐릭터 복장·얼굴 일치가 우수."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "'황봉업 외 인물 추가 금지' 규정을 어기고 전경에 슬링을 한 인물을 넣은 하드 위반이며, 배 위 인물의 복장도 레퍼런스(남색 니트+고무 앞치마)와 어긋나고 인물이 너무 작아 '불만 가득한 표정'이 읽히지 않음."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "황봉업은 배 갑판 위에서 카메라(부두 쪽) 정면을 향해 서 있고 시선은 부두 눈높이 카메라 방향, 즉 대치 상대가 있을 왼쪽 부두 쪽을 향한다. 배의 뱃머리는 화면 왼쪽 아래(부두 쪽)를 향하고 부두-배 사이의 물 간격이 화면 중앙~왼쪽에 보인다. 인물은 화면 중앙 우측에 배치되어 지시와 일치.",
      "built_space": "낚시 어선 한 척: 조타실 1개, 안테나/레이더 마스트, 배기 연통, 그물 릴, 계류 로프와 노란 펜더가 부두 볼라드에 걸려 있다. 부두는 참조 사진과 동일한 콘크리트 방파제, 낮은 연석, 낡은 볼라드, 배경의 붉은 등대와 바위 방파제, 섬 능선이 일치. 중복된 구조물이나 여분의 조타실 없음.",
      "entities": "인물은 1명뿐 — 50대 한국 남성, 검은 머리, 넓은 얼굴, 남색 니트 위에 낡은 고무 앞치마로 캐릭터 레퍼런스와 복장·얼굴이 잘 일치. 표정은 웃음기 없이 굳어 불만스러운 인상. 프롬프트가 금지한 추가 인물 없음.",
      "hard_violations": [],
      "physics": "두 발이 갑판 위에 확실히 놓여 꼿꼿하게 직립, 팔은 몸 옆으로 자연스럽게 내려져 있음. 배는 물에 떠 있고 로프·펜더로 부두에 고정되어 지지 관계가 명확. 공중에 떠 있는 물체 없음."
     },
     {
      "label": "A",
      "direction": "왼쪽 부두 위 인물(팔걸이를 한 젊은 남성)은 배 쪽을 바라보고, 배 위 남성은 부두 쪽을 향해 서 있다. 카메라는 부두 눈높이 광각. 그러나 프롬프트가 명시한 '황봉업만' 프레임에 들어가야 하는 지시를 어기고 인우로 보이는 인물이 전경에 추가됨.",
      "built_space": "어선 1척: 조타실 1개, 집어등 줄, 그물 릴, 펜더, 계류 로프 — 구조 자체는 정합. 부두는 참조와 유사하나 붉은 등대가 매우 작게 좌측 멀리 배치. 배 위 인물은 조타실 옆 갑판에 서 있음.",
      "entities": "인물 2명: (1) 부두의 검은 재킷·팔 슬링 착용 청년 — 프롬프트가 '절대 추가하지 말라'고 한 인물. (2) 배 위 남성은 국방색 작업복 차림으로 캐릭터 레퍼런스의 남색 니트+고무 앞치마 복장과 불일치하고 얼굴 크기가 작아 동일 인물 확인이 어려움. 또한 화면 중앙 우측이 아니라 인물이 중앙보다 우측 조타실 옆이지만 크기가 매우 작음.",
      "hard_violations": [
       "프롬프트가 명시적으로 금지한 추가 인물(인우로 보이는 팔 깁스 남성)이 프레임 안에 등장 — invented/extra person"
      ],
      "physics": "두 인물 모두 두 발로 지면(부두/갑판)에 지지, 부양 문제 없음. 배는 로프와 펜더로 부두에 계류되어 있음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "readings": [
   {
    "label": "A",
    "direction": "카메라는 선착장 위 눈높이에서 배를 향하고 있으나, 프레임 좌측 전경에 팔걸이(슬링)를 한 젊은 남자가 뒷모습으로 서서 배 위 남자를 바라보고 있다. 배 위 남자(황 씨로 추정)는 조타실 옆에 서서 선착장 쪽(좌측 전방)을 바라본다. 시선 방향 자체는 서로 맞물리지만, 샷 텍스트는 황 씨 한 사람만을 지시한다.",
    "built_space": "낚시 어선의 조타실, 난간, 집어등 줄, 그물 릴 등이 배치되어 있고 선착장은 좌측에서 뒤로 이어진다. 레퍼런스의 콘크리트 방파제·계선주·타이어 펜더·원경 섬은 어느 정도 반영되었으나 붉은 등대가 매우 작게 좌측 끝에 흐릿하게 존재. 배는 선착장 우측에 정박, 접안면이 카메라를 향해 보인다 — 구조 자체는 성립.",
    "entities": "황 씨: 50대 중반 한국 남성으로 보이나 캐릭터 레퍼런스의 남색 니트+고무 앞치마+장화 복장이 아니라 회녹색 작업복 상하의를 입었고, 얼굴 유사도도 낮은 편(더 젊고 마른 인상). 추가로 캐스트/슬링을 한 인우 캐릭터가 실제 인물로 프레임에 등장 — 프롬프트는 '샷 텍스트가 보여주지 않는 인물은 절대 추가하지 말 것'이며 등장 가능 인물은 황봉업 한 명뿐이라고 못박음.",
    "hard_violations": [
     "샷 텍스트에 없는 인물(팔 슬링을 한 남성)을 전경에 추가 — PEOPLE 조항 '황봉업 외 누구도 안 됨' 위반",
     "황 씨가 화면 중앙-우측이 아니라 중앙 부근이며, 추가 인물이 좌측 여백(선착장-배 간극)을 점유해 지정된 구도 논리 훼손"
    ],
    "physics": "두 인물 모두 발이 각각 선착장 콘크리트와 갑판에 닿아 체중을 지지한다. 배는 물에 떠 있고 계류 로프와 펜더가 접안 상태를 지지. 부유하거나 지지 없는 물체는 없음."
   },
   {
    "label": "B",
    "direction": "카메라는 선착장 위 서 있는 눈높이에서 짧은 간격 너머 배를 정면으로 바라본다. 황 씨는 갑판 위에 서서 시선을 카메라/선착장 쪽(좌측 약간 아래)으로 곧게 향하고 있으며, 불만 섞인 굳은 표정. 시선 대상은 선착장에 선 상대(프레임 밖) 방향으로 정확히 떨어진다.",
    "built_space": "낚시 어선의 조타실, 안테나 마스트, 배기 연통, 그물더미, 릴이 있고 선체 접안면이 카메라를 향한다. 선착장은 좌측 하단 전경에서 뒤쪽 배 쪽으로 길게 뻗어 접근선을 만들고, 계선주·타이어와 노란 펜더·붉은 등대·원경 섬·바위 방파제 등 레퍼런스 고정 요소가 잘 일치. 조타실은 하나, 구조 중복 없음.",
    "entities": "황 씨: 한국인 50대 남성, 짙은 머리, 넓은 얼굴형으로 캐릭터 레퍼런스와 상당히 근접. 복장도 레퍼런스의 남색 니트 + 낡은 고무 앞치마 + 어두운 바지/장화를 그대로 재현. 프레임 내 다른 인물 없음 — 샷 텍스트 요구와 일치.",
    "hard_violations": [],
    "physics": "황 씨는 두 발로 갑판 위에 서 있고 무게가 다리에 실려 있다. 배는 계류 로프와 펜더로 부두에 접안, 물 위에 정상적으로 떠 있음. 공중에 지지 없이 뜬 물체 없음."
   }
  ],
  "totals": {
   "A": 6,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "샷 텍스트가 지시하지 않은 슬링 착용 남성을 전경에 추가해 'PEOPLE — 황봉업 외 절대 금지' 조항을 정면 위반했고, 황 씨 의상·얼굴도 캐릭터 레퍼런스와 어긋난다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "황 씨 단독으로 갑판 중앙-우측에 꼿꼿이 서서 선착장 쪽을 노려보고, 좌측에 부두-배 간극과 접근선을 남긴 와이드 구도·대낮 광선·레퍼런스 복장/장소 요소가 모두 부합한다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_mainland_harbor_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 황봉업(황 씨): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:950885>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "어선 조타실 지붕의 하얀색 레이더 장비에 'WAFA'라는 텍스트가 생성되었습니다 (텍스트 생성 금지 위반).",
     "fix_en": "Remove the text 'WAFA' from the white radar dome on the roof of the boat cabin, replacing it with a plain white surface. Preserve the man, his pose and expression, his clothing, the boat, the pier, the background landscape, and the lighting exactly as they are.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the text 'WAFA' from the white radar dome on the roof of the boat cabin, replacing it with a plain white surface. Preserve the man, his pose and expression, his clothing, the boat, the pier, the background landscape, and the lighting exactly as they are.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·인물 배치·조명은 계약대로지만 조타실 상단에 판독 가능한 문자 라벨이 남아 '어떤 텍스트도 금지' 조항을 정면으로 위반한다.  ★위반: [claude-opus] 조타실 상부 흰 패널에 판독 가능한 문자열이 렌더링됨 — 'No text, captions, watermarks or annotations anywhere' 위반 / [gpt] 선실 상단 흰색 장비에 ‘HAFA’라는 불필요한 문자가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "A와 거의 동일한 구도·인물·소품·광선을 유지한 채 문자 라벨만 제거했고, 얼굴이 참조에 조금 더 근접해 손실 없이 결함만 수리한 쪽이다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.417,
      "B": 2.0
     },
     "adjusted": {
      "A": 0.917,
      "B": 2.0
     },
     "violations": {
      "A": [
       "[claude-opus] 조타실 상부 흰 패널에 판독 가능한 문자열이 렌더링됨 — 'No text, captions, watermarks or annotations anywhere' 위반",
       "[gpt] 선실 상단 흰색 장비에 ‘HAFA’라는 불필요한 문자가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·인물 배치·조명은 계약대로지만 조타실 상단에 판독 가능한 문자 라벨이 남아 '어떤 텍스트도 금지' 조항을 정면으로 위반한다.  ★위반: [claude-opus] 조타실 상부 흰 패널에 판독 가능한 문자열이 렌더링됨 — 'No text, captions, watermarks or annotations anywhere' 위반 / [gpt] 선실 상단 흰색 장비에 ‘HAFA’라는 불필요한 문자가 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "A와 거의 동일한 구도·인물·소품·광선을 유지한 채 문자 라벨만 제거했고, 얼굴이 참조에 조금 더 근접해 손실 없이 결함만 수리한 쪽이다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "부두 눈높이 와이드, 중앙우측 갑판의 꼿꼿한 황 씨, 좌측 부두-배 간격 노출까지 지시대로 충족하고 인물·로케이션 레퍼런스 일치, 금지 문자 없음."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "프레이밍·인물은 A와 거의 동일하나 레이더 돔에 'HAFA' 문자가 새로 생겨 텍스트 금지 조항을 어겼고, 갑판의 붉은 상자 소품이 이유 없이 사라져 계약이 요구하지 않은 손실만 남았다.  ★위반: [claude-opus] 조타실 위 장비 돔에 'HAFA' 형태의 판독 가능한 문자 표기 — '텍스트·캡션·워터마크 금지' 위반 / [gpt] 조타실 지붕의 흰 장비에 ‘HAFA’로 보이는 불필요한 문자가 노출되어 ‘문자·캡션·워터마크 금지’ 조건을 위반한다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.417
     },
     "adjusted": {
      "A": 2.0,
      "B": 0.917
     },
     "violations": {
      "B": [
       "[claude-opus] 조타실 위 장비 돔에 'HAFA' 형태의 판독 가능한 문자 표기 — '텍스트·캡션·워터마크 금지' 위반",
       "[gpt] 조타실 지붕의 흰 장비에 ‘HAFA’로 보이는 불필요한 문자가 노출되어 ‘문자·캡션·워터마크 금지’ 조건을 위반한다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "부두 눈높이 와이드, 중앙우측 갑판의 꼿꼿한 황 씨, 좌측 부두-배 간격 노출까지 지시대로 충족하고 인물·로케이션 레퍼런스 일치, 금지 문자 없음."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·인물은 A와 거의 동일하나 레이더 돔에 'HAFA' 문자가 새로 생겨 텍스트 금지 조항을 어겼고, 갑판의 붉은 상자 소품이 이유 없이 사라져 계약이 요구하지 않은 손실만 남았다.  ★위반: [claude-opus] 조타실 위 장비 돔에 'HAFA' 형태의 판독 가능한 문자 표기 — '텍스트·캡션·워터마크 금지' 위반 / [gpt] 조타실 지붕의 흰 장비에 ‘HAFA’로 보이는 불필요한 문자가 노출되어 ‘문자·캡션·워터마크 금지’ 조건을 위반한다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S53sh1__bgfirst_bg.png",
   "bg_asset_id": "fec13fda-50d3-4945-9920-0a88df918669",
   "bg_record_key": "S53sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S60sh7": {
  "input_fingerprint": "b6756f0f94c8d22f",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 갯바위 쪽을 향해 충격에 휩싸여 눈이 휘둥그레진 채 굳어 있는 김지후(강수리영)와 정인우의 얼굴.\n\nLOCATION (lock): The exposed coastal rock path directly above the body lodged in the surf. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: On the reverse cut, hold an eye-level tight frontal two-shot less than a meter away, placing 김지후(강수리영) and 정인우 shoulder-to-shoulder with both widened gazes passing just beyond the lens toward the corpse. Keep their faces equally prominent but slightly offset rather than symmetrically centered, preserving the shared eyeline while allowing 김지후(강수리영)'s paling terror to register beside 정인우's stunned reaction.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: coastal path (behind the two people); used as Provides a narrow, defocused trace of the coastal path behind the reaction; coastal rocks (partially visible in the background); used as Keeps the pair situated along the rocky coast without competing with their faces.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime coastal light is held in high facial contrast, emphasizing widened eyes and 김지후(강수리영)'s visibly paling fear without adding a new source.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same shoreline, body position, rocks, waves, and daylight. Exclude a body-only establishing composition; show both witnesses frozen in shock while looking toward it.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The corpse remains wedged among the rocks; Jihu is still without her necklace and Inwoo still has one arm casted.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face); 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 9,
   "C": 6
  },
  "selected": "B",
  "ranking": [
   "B",
   "A",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "정확한 외화면 시선과 정인우의 깁스·슬링은 좋지만, 김지후의 모자와 정인우의 버건디색 상의가 빠지고 여성의 얼굴 일치도도 낮다."
   },
   {
    "label": "B",
    "score": 9,
    "verdict_ko": "눈높이 타이트 정면 투샷, 공유된 갯바위 쪽 시선, 충격 표정, 동일 해안 배경과 정인우의 깁스 상태를 가장 충실히 구현했으며 김지후의 모자 누락만 두드러진다."
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "구도와 김지후의 모자는 정확하지만 한쪽 팔 깁스를 정인우가 아닌 김지후에게 배정해 명시된 지속 상태를 크게 어겼다.  ★위반: [gpt] 정인우에게 있어야 할 한쪽 팔 깁스를 김지후에게 부여해, 김지후에게 지시되지 않은 깁스 물체를 추가하고 인물별 고정 상태를 뒤바꿨다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S60sh6_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:813104>"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:787232>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "[Critical] 화면 우측 가장자리에 프롬프트에서 허용하지 않은 인물(시체)의 다리가 나타나 있으며, 두 사람의 시선 방향(카메라 렌즈 너머)을 고려할 때 공간적으로도 모순됩니다.",
     "fix_en": "Erase the blue-jeaned legs protruding into the frame on the far right edge, filling the space with the defocused dark coastal rocks and grey sea to match the surrounding background. Preserve the two characters' faces, expressions, poses, clothing, the cast on the man's arm, the lighting, and the rest of the background exactly as they are."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Erase the blue-jeaned legs protruding into the frame on the far right edge, filling the space with the defocused dark coastal rocks and grey sea to match the surrounding background. Preserve the two characters' faces, expressions, poses, clothing, the cast on the man's arm, the lighting, and the rest of the background exactly as they are.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "눈높이 정면 타이트 투샷, 렌즈를 스치는 공유 아이라인, 슬링 캐스트·목걸이 없음 계약을 모두 충족하고 머리카락·모직·콘크리트의 물성이 살아 있어 재질 리얼리즘 요구에 더 부합한다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "구도·시선·의상은 A와 동일하나 계약이 요구하지 않은 전반적 스무딩으로 머리카락 결과 배경 질감이 뭉개져 '실제 물성' 조항에서 손해만 발생했다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.875
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.875
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "눈높이 정면 타이트 투샷, 렌즈를 스치는 공유 아이라인, 슬링 캐스트·목걸이 없음 계약을 모두 충족하고 머리카락·모직·콘크리트의 물성이 살아 있어 재질 리얼리즘 요구에 더 부합한다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "구도·시선·의상은 A와 동일하나 계약이 요구하지 않은 전반적 스무딩으로 머리카락 결과 배경 질감이 뭉개져 '실제 물성' 조항에서 손해만 발생했다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "동일한 로케이션·의상·깁스 유지에 하드 위반은 없으나, 시선이 거의 렌즈 직시로 '렌즈를 살짝 지나 갯바위 쪽'이라는 아이라인 지시에서 벗어나고, 김지후의 얼굴형·머리 흐트러짐과 정인우의 굵어진 눈썹 등 계약이 요구하지 않은 인물 동일성 변화가 있으며 배경이 덜 디포커스되어 얼굴과 경쟁한다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "같은 해안 갯바위·주광·의상·목걸이 없음·한팔 깁스를 모두 유지한 채, 두 시선이 렌즈를 비껴 시신 쪽으로 향하는 공유 아이라인을 더 정확히 실현하고 배경을 좁고 흐릿한 해안길 흔적으로 눌러 얼굴 클로즈업 지시에 부합하며, 두 인물의 얼굴이 캐릭터 레퍼런스에 더 근접한다.  ★위반: [gpt] 우측 하단에 샷 텍스트가 보이도록 지정하지 않은 제3의 신체로 읽히는 형상이 추가되어 있으며, 시신이라면 두 증인의 시선 목표와 반대편인 뒤쪽에 잘못 배치되어 있다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.875,
      "B": 1.444
     },
     "adjusted": {
      "A": 1.875,
      "B": 1.194
     },
     "violations": {
      "B": [
       "[gpt] 우측 하단에 샷 텍스트가 보이도록 지정하지 않은 제3의 신체로 읽히는 형상이 추가되어 있으며, 시신이라면 두 증인의 시선 목표와 반대편인 뒤쪽에 잘못 배치되어 있다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "A"
     },
     "agreed": false
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "동일한 로케이션·의상·깁스 유지에 하드 위반은 없으나, 시선이 거의 렌즈 직시로 '렌즈를 살짝 지나 갯바위 쪽'이라는 아이라인 지시에서 벗어나고, 김지후의 얼굴형·머리 흐트러짐과 정인우의 굵어진 눈썹 등 계약이 요구하지 않은 인물 동일성 변화가 있으며 배경이 덜 디포커스되어 얼굴과 경쟁한다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "같은 해안 갯바위·주광·의상·목걸이 없음·한팔 깁스를 모두 유지한 채, 두 시선이 렌즈를 비껴 시신 쪽으로 향하는 공유 아이라인을 더 정확히 실현하고 배경을 좁고 흐릿한 해안길 흔적으로 눌러 얼굴 클로즈업 지시에 부합하며, 두 인물의 얼굴이 캐릭터 레퍼런스에 더 근접한다.  ★위반: [gpt] 우측 하단에 샷 텍스트가 보이도록 지정하지 않은 제3의 신체로 읽히는 형상이 추가되어 있으며, 시신이라면 두 증인의 시선 목표와 반대편인 뒤쪽에 잘못 배치되어 있다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 14
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S60sh6"
  }
 },
 "S62sh8": {
  "input_fingerprint": "cc82c6fe9cacbd4d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): early evening.\n\nSHOT TEXT (authoritative, Korean): 황봉업(황 씨)이 김여상의 시체를 허리 높이로 번쩍 들어 올린 채 자신의 배 쪽으로 몸을 반쯤 튼 순간.\n\nLOCATION (lock): The narrow gap between the police patrol boat and a fishing boat drawn alongside on the evening sea. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From waist height beside the adjacent boat sides, hold a wide lateral view as 황봉업(황 씨) supports 김여상’s body at his waist and freezes halfway through his turn toward the fishing boat on screen right. Keep both boats’ edges visible as scale references, with 황 씨 and the corpse occupying the central midground rather than allowing either vessel to overwhelm the frame.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 황봉업(황 씨) in the middle-center of the frame, midground, moves toward 황 씨의 낚시 어선; 김여상 in the middle-left of the frame, midground; 황 씨의 낚시 어선 in the middle-right of the frame, background.\n- KEY BACKGROUND ELEMENTS: 경찰 순찰정 측면 (Adjacent to the fishing boat) — Its outward side faces the neighboring fishing boat across the gap; used as Forms the left boundary and starting point of the transfer; 황 씨의 낚시 어선 측면 (Adjacent to the patrol boat) — Its receiving side faces the patrol boat; used as Marks the destination on the right side of the transfer path.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key evening illumination silhouettes the transfer against the dark sea while preserving readable detail in the two figures.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same patrol-boat deck, darkening sea, stopped-engine stillness, corpse, and early-evening light. Exclude the officer's looting action; show the boatman lifting the corpse toward his own vessel.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): His swollen corpse hangs completely limp in 황봉업(황 씨)’s hold at waist height as it is carried toward the fishing boat. The unsupported head and limbs droop downward while the torso is borne across the carrier’s arms.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Yeosang’s corpse, stripped of the valuables Changmo took, is being transferred intact to Hwang’s fishing boat.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 황봉업(황 씨) right now, so 황봉업(황 씨)'s hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 황봉업(황 씨): its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 황봉업(황 씨) (Korean man in his mid-50s, dark graying hair, broad genial face); 김여상 (Korean man in his mid-40s, dark hair, ordinary mature face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "황 씨는 순찰정 갑판 위 중앙에 서서 시신을 허리 높이로 안고 화면 오른쪽(어선 쪽)으로 몸을 반쯤 튼 상태다. 그의 시선은 오른쪽 아래, 시신과 어선 방향을 향한다. 시신의 머리는 화면 왼쪽으로 젖혀져 있고 다리는 오른쪽 난간 너머 어선 쪽으로 향해 이동 방향과 일치한다. 김여상은 중앙-좌측, 황 씨는 중앙, 어선은 중앙-우측 배경으로 프레임 레이아웃 지시와 일치.",
    "built_space": "왼쪽에 순찰정 조타실과 초록색 갑판·난간(이전 스틸의 초록 갑판, 흰 상부구조, 로프와 일치), 오른쪽에 흰·파란 어선 선측과 조타실이 있고 그 사이에 좁은 바닷물 틈이 보인다. 두 선박의 가장자리가 모두 보이며 어느 쪽도 프레임을 압도하지 않는다. 카메라는 인접한 선측 옆 허리 높이 수평 시점으로 읽힌다. 구조물 중복 없음.",
    "entities": "황 씨: 50대 한국 남성, 검은 머리에 넓고 온화한 얼굴 — 이전 스틸의 감청색 셔츠 차림과 일치(캐릭터 레퍼런스의 앞치마·장화는 아님). 김여상: 40대 남성, 이전 스틸에서 입고 있던 어두운 올리브 상의와 검은 바지 그대로, 창백한 얼굴. 인물은 두 명뿐, 추가 인물·텍스트 없음.",
    "hard_violations": [],
    "physics": "시신은 황 씨의 두 팔에 확실히 받쳐져 있다 — 오른팔이 등/겨드랑이를, 왼손이 무릎 뒤를 잡고 있어 접촉점이 명확하다. 머리·왼팔은 중력대로 아래로 늘어지고 손가락도 이완돼 있다. 황 씨는 갑판에 두 발로 서서 무게를 견디는 자세이며 상체가 약간 뒤로 기울어 하중 보상이 보인다. 공중에 떠 있는 것은 없음."
   },
   {
    "label": "B",
    "direction": "황 씨는 오른쪽 목선 안에서 몸을 좁은 틈 위로 내밀어 시신을 왼쪽(순찰정) 쪽으로 뻗고 있다. 즉 이동 방향이 어선→순찰정 또는 시신이 물 위에 매달린 상태로, 프롬프트가 요구한 '어선 쪽(화면 오른쪽)으로 반쯤 트는' 방향과 반대로 읽힌다. 시선은 시신 얼굴을 향한다. 김여상은 중앙-좌측이지만 두 배 사이 물 위 허공에 있다.",
    "built_space": "왼쪽 순찰정(초록 갑판, 구명환, 난간)과 오른쪽 목조 어선 선측이 모두 보이고 좁은 틈도 존재해 장소 자체는 맞다. 다만 오른쪽 어선의 선체와 조타실이 프레임 우측을 크게 차지하고, 황 씨가 서 있는 발판이 보이지 않는다.",
    "entities": "황 씨: 중년 한국 남성, 어두운 작업복 — 얼굴은 레퍼런스와 대체로 유사하나 이전 스틸의 감청색 셔츠와 다르다. 김여상: 어두운 상의의 성인 남성, 창백. 추가 인물 없음. 텍스트/워터마크 없음.",
    "hard_violations": [
     "시신이 두 배 사이 바닷물 위 허공에 가로로 떠 있고, 황 씨의 두 팔은 몸통 옆을 얕게 감싸고 있을 뿐 다리·하체를 받치는 접촉점이 전혀 없어 실제로 이 자세를 지탱할 지지대가 없다(중력상 불가능한 부양).",
     "황 씨 본인의 하체와 발이 어디에도 딛고 있지 않은 채 선측 위로 상체가 크게 돌출돼 있어 이 하중을 지탱할 지지가 보이지 않는다."
    ],
    "physics": "시신은 허리 높이가 아니라 물 위에 수평으로 매달려 있고, 팔 아래로 받침이 없다. 황 씨의 팔은 겨드랑이 부근만 스치듯 닿아 있어 성인 남성의 전체 무게를 지탱할 수 없다. 황 씨 자신도 뱃전에 상체를 걸친 형태로 하중 균형이 성립하지 않는다."
   },
   {
    "label": "C",
    "direction": "황 씨는 중앙에서 시신을 허리 높이로 안고 있으며 시선과 상체가 화면 오른쪽 어선 쪽을 향해 반쯤 틀어져 있어 방향은 맞다. 시신의 다리는 오른쪽 어선 쪽을 향한다. 레이아웃(황 씨 중앙, 시신 좌측, 어선 우측)도 지시와 부합.",
    "built_space": "왼쪽 순찰정 선측, 오른쪽 목조 어선, 가운데 좁은 물 틈은 있으나 황 씨가 그 물 틈 한가운데에 서 있다 — 즉 갑판이 아니라 수면 위에 서 있는 것처럼 보인다. 발목까지 물에 잠긴 듯한 위치라 두 선박 사이 바다에서 사람이 설 수 있는 지지면이 존재하지 않는다.",
    "entities": "황 씨: 50대, 희끗한 머리, 넓은 얼굴, 캐릭터 레퍼런스의 작업복+가죽 앞치마+장화 착용 — 캐릭터 레퍼런스와는 잘 맞으나 이전 스틸의 감청색 셔츠 상태와는 어긋난다. 김여상: 40대 남성, 회색 티셔츠와 검은 바지, 창백한 얼굴. 인물 2명, 텍스트 없음.",
    "hard_violations": [
     "황 씨가 두 선박 사이 바다 수면 위에 서 있어 그를 받치는 갑판·지지면이 전혀 보이지 않는다(물 위에 선 상태 — 지지 없음)."
    ],
    "physics": "시신 자체는 두 팔(등 아래·무릎 뒤)에 제대로 받쳐지고 머리와 왼팔이 중력대로 늘어져 있어 시신 지지는 양호하다. 그러나 운반자 본인의 발 아래는 물뿐이며, 갑판 가장자리도 그의 위치보다 훨씬 뒤에 있어 지탱하는 표면을 찾을 수 없다."
   }
  ],
  "totals": {
   "A": 8,
   "C": 5,
   "B": 3
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "허리 높이로 시신을 확실히 받쳐 든 채 오른쪽 어선 쪽으로 반쯤 몸을 튼 순간을 정확히 포착했고, 순찰정 갑판·초록 바닥·좁은 틈 등 이전 스틸의 장소와 복장이 유지되며 두 선박 가장자리가 모두 스케일 기준으로 보이는 와이드 구도라 지시에 가장 충실하다."
   },
   {
    "label": "C",
    "score": 5,
    "verdict_ko": "시신의 늘어진 자세와 어선 쪽 회전 방향은 맞지만, 황 씨가 두 배 사이 바닷물 위에 발 디딜 곳 없이 서 있어 지지 불가능한 스테이징이며 이전 스틸의 감청색 셔츠 대신 앞치마 차림으로 연속성도 어긋난다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "시신이 두 배 사이 수면 위에 하체 받침 없이 수평으로 떠 있고 운반자 자신도 뱃전에 상체만 걸쳐 지지가 성립하지 않으며, 이동 방향도 어선 쪽이 아닌 순찰정 쪽으로 읽혀 샷 텍스트의 순간과 반대다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S62sh4_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 황봉업(황 씨): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:950885>"
   },
   {
    "label": "CHARACTER REFERENCE — 김여상: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:552293>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "시체를 든 인물이 황 씨(스웨터, 고무 앞치마)가 아닌 이전 컷의 경찰관(남색 셔츠) 얼굴과 복장으로 잘못 생성되었습니다.",
     "fix_en": "Replace the head and clothing of the standing man with Hwang's broad face, worn dark blue sweater, and black rubber apron from his character reference. Preserve the people present and their positions, the corpse's pose and clothing, the set, the light, and the framing.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "시체의 오른팔이 운반자의 등을 감싸고 왼손이 배 위에 있으며 종아리가 수평으로 들려 있어, 축 늘어져야 하는 시체의 중력 규칙을 크게 위반합니다.",
     "fix_en": "Redraw the corpse's right arm, left arm, and lower legs to hang completely limp straight downward toward the deck, removing them from the carrier's back and stomach; fill the exposed areas with the carrier's torso and the boat background. Preserve the people present and their positions, the men's clothing, the set, the light, and the framing.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the head and clothing of the standing man with Hwang's broad face, worn dark blue sweater, and black rubber apron from his character reference. Preserve the people present and their positions, the corpse's pose and clothing, the set, the light, and the framing.\n- Redraw the corpse's right arm, left arm, and lower legs to hang completely limp straight downward toward the deck, removing them from the carrier's back and stomach; fill the exposed areas with the carrier's torso and the boat background. Preserve the people present and their positions, the men's clothing, the set, the light, and the framing.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "구도·지향·물리 지지 모두 계약대로지만, 황 씨가 캐릭터 레퍼런스가 고정한 남색 니트+고무 앞치마 어부 복장이 아닌 셔츠 차림이라 의상 락에서 손해를 본다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "동일한 와이드 측면 구도·좌(순찰정)/우(어선) 배치·허리 높이 운반과 중력 순응 시신을 유지하면서, 황 씨의 앞치마·니트 착장까지 캐릭터 레퍼런스대로 복원해 다른 요소는 거의 건드리지 않았다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.556,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.556,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "구도·지향·물리 지지 모두 계약대로지만, 황 씨가 캐릭터 레퍼런스가 고정한 남색 니트+고무 앞치마 어부 복장이 아닌 셔츠 차림이라 의상 락에서 손해를 본다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "동일한 와이드 측면 구도·좌(순찰정)/우(어선) 배치·허리 높이 운반과 중력 순응 시신을 유지하면서, 황 씨의 앞치마·니트 착장까지 캐릭터 레퍼런스대로 복원해 다른 요소는 거의 건드리지 않았다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "와이드 측면 구도, 좌우 두 선체 경계, 허리 높이 운반과 중력에 순응한 시체 자세를 모두 충족하며, 황 씨의 니트+고무 앞치마 복장이 캐릭터 레퍼런스와 정확히 일치한다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "구도·방향·물리는 A와 사실상 동일하나, 계약이 요구하지 않은 변경으로 황 씨의 고무 앞치마와 니트가 사라져 캐릭터 레퍼런스 의상 락이 깨졌고 갑판 로프 등 세트 요소도 일부 소실됐다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.528
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.528
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "와이드 측면 구도, 좌우 두 선체 경계, 허리 높이 운반과 중력에 순응한 시체 자세를 모두 충족하며, 황 씨의 니트+고무 앞치마 복장이 캐릭터 레퍼런스와 정확히 일치한다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "구도·방향·물리는 A와 사실상 동일하나, 계약이 요구하지 않은 변경으로 황 씨의 고무 앞치마와 니트가 사라져 캐릭터 레퍼런스 의상 락이 깨졌고 갑판 로프 등 세트 요소도 일부 소실됐다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 13,
     "B": 17
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S62sh4"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S63sh13": {
  "input_fingerprint": "a83cbd9fa7428f8c",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 피범벅이 된 자신의 두 손을 내려다보며 극심한 충격으로 두 눈을 부릅뜬 김지후(강수리영)의 얼굴 클로즈업.\n\nLOCATION (lock): The dining-table area inside the restaurant beside platters of raw meat. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Complete the upward tilt directly in front of 김지후(강수리영) at eye level, ending on the explicitly extreme facial close-up as her widened eyes remain aimed down toward her blood-covered hands. Preserve the tops of both raised hands along the lower edge so the cause of her shock stays inside the frame, with no other person intruding.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 김지후(강수리영) in the middle-center of the frame, foreground.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key interior contrast fixes attention on her widened eyes and the deep red blood at the lower frame edge.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same raw-meat table, nighttime restaurant lighting, blood-smeared hands and mouth, and seated position. Exclude the other diners from the tight frame; focus on her horrified reaction to her own hands.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu’s mouth and both hands remain heavily smeared with raw blood, and her necklace remains missing. Inwoo still wears his cast.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 5,
   "B": 7,
   "C": 9
  },
  "selected": "C",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "시선이 손이 아닌 정면을 향하고, 배경에 생고기 식탁 등 잠금 요소가 거의 보이지 않으며 상의 색도 이전 스틸과 달라 장소·연속성 충실도가 가장 낮음."
   },
   {
    "label": "B",
    "score": 7,
    "verdict_ko": "장소 잠금(창·고기 접시)과 의상·피 연속성은 좋으나, 시선이 손을 내려다보지 않고 정면을 응시해 샷 텍스트의 핵심 동작에서 C에 밀림."
   },
   {
    "label": "C",
    "score": 9,
    "verdict_ko": "부릅뜬 눈이 아래 손 쪽으로 향한 상태로 틸트 완료 클로즈업을 실현했고, 하단에 피범벅 두 손이 유지되며 생고기 식탁·야간 조명·의상·목걸이 부재까지 모두 일치."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S63sh8_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:813104>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "캐릭터가 카메라 렌즈를 정면으로 응시하고 있습니다. 프롬프트에 따라 시선이 아래쪽의 피 묻은 손을 향해 내려다보아야 합니다.",
     "fix_en": "Redraw only her eyeballs so the irises and pupils are looking sharply downwards toward her hands at the bottom of the frame. Preserve the widened shape of her eyelids, her face, the blood on her skin and hands, her exact hand positions, her clothing, the lighting, and the background exactly as they are.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw only her eyeballs so the irises and pupils are looking sharply downwards toward her hands at the bottom of the frame. Preserve the widened shape of her eyelids, her face, the blood on her skin and hands, her exact hand positions, her clothing, the lighting, and the background exactly as they are.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "시선이 정면 카메라를 향해 크게 부릅뜬 채 고정되어 있다. 눈동자가 렌즈를 똑바로 응시하며, 아래쪽 손을 내려다보는 방향성이 전혀 없다. 샷 텍스트가 요구한 '자신의 두 손을 내려다보는' 시선이 성립하지 않는다. 두 손은 프레임 하단에서 손바닥을 카메라 쪽으로 펼쳐 얼굴 양옆에 올라와 있고, 그 자체는 화면 안에 남아 있다.",
      "built_space": "배경은 참고 스틸과 같은 고깃집 식탁: 창틀(밤 바다), 스테인리스 물통, 생고기 접시 두 개, 오른쪽 구이용 그릴, 반찬 그릇, 나무 의자 등받이. 개수와 배치가 참고 스틸의 공간과 정합하며 다른 손님은 프레임에서 제외되어 있다. 배경 요소가 과대하게 확대된 곳은 없다.",
      "entities": "20대 한국 여성, 긴 검은 머리(흐트러진 잔머리), 참고 인물과 유사한 얼굴형. 입가와 두 손이 붉은 생피로 범벅. 목걸이 없음(계약대로). 상의는 이전 스틸과 같은 베이지-회색 니트. 다만 눈이 홍채 주변 흰자가 과도하게 드러나 만화적으로 확대된 인상, 눈 크기가 부자연스럽게 커져 인물 동일성이 다소 흔들린다.",
      "hard_violations": [],
      "physics": "두 손은 팔뚝으로 지지되어 얼굴 앞에 들려 있고 손목-팔의 연결이 보인다. 상체는 식탁 앞에 앉은 자세로 무게가 실려 있다. 공중에 떠 있는 물체 없음. 피는 손가락과 손등에 젖은 광택으로 흐르고 손등 아래로 뭉침이 보여 액체로 읽힌다."
     },
     {
      "label": "B",
      "direction": "시선이 정면보다 살짝 아래로 내려가 있고 눈동자가 아래 방향(들어올린 두 손 쪽)으로 향해 있어, '자신의 피범벅 손을 내려다본다'는 샷 텍스트의 시선 타깃이 성립한다. 두 손은 프레임 하단에서 얼굴 양옆으로 올라와 상단이 잘리지 않고 유지된다.",
      "built_space": "A와 동일한 고깃집 식탁 공간: 창틀·스테인리스 물통·생고기 접시 2개·오른쪽 그릴·반찬 그릇·의자 등받이. 참고 스틸의 재질과 조명(어두운 실내, 따뜻한 저조도)과 일치. 다른 인물 없음, 배경 스케일 정상.",
      "entities": "20대 한국 여성, 긴 검은 머리, 참고 인물과 일치하는 얼굴 비례. 눈은 놀람으로 크게 떴으나 홍채·공막 비율이 해부학적으로 정상 범위이며, 입가와 두 손의 선혈, 목걸이 부재 모두 계약 충족. 의상은 이전 스틸과 동일한 니트.",
      "hard_violations": [],
      "physics": "두 손은 팔로 받쳐져 얼굴 앞에 들려 있고, 상체는 착석 자세. 무지지 부유물 없음. 피는 손가락 능선을 따라 젖은 반사와 흐름을 보이며 스티커처럼 붙어 있지 않다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "공간·소품·핏자국은 계약대로지만, 눈동자가 렌즈를 정면 응시해 '두 손을 내려다본다'는 샷 텍스트의 핵심 시선이 무너졌고, 안구가 과장되게 확대되어 '눈은 항상 해부학적으로 정상' 조항과 인물 동일성에서 손해를 봤다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 배경·의상·혈흔을 그대로 유지한 채 시선만 아래(자신의 두 손)로 내리고 안구 비례를 정상 범위로 되돌려, 샷 텍스트의 '피범벅 손을 내려다보며 부릅뜬 눈'과 해부학적 정상 조항을 함께 충족한 최소 변경 수정본이다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "공간·소품·핏자국은 계약대로지만, 눈동자가 렌즈를 정면 응시해 '두 손을 내려다본다'는 샷 텍스트의 핵심 시선이 무너졌고, 안구가 과장되게 확대되어 '눈은 항상 해부학적으로 정상' 조항과 인물 동일성에서 손해를 봤다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 배경·의상·혈흔을 그대로 유지한 채 시선만 아래(자신의 두 손)로 내리고 안구 비례를 정상 범위로 되돌려, 샷 텍스트의 '피범벅 손을 내려다보며 부릅뜬 눈'과 해부학적 정상 조항을 함께 충족한 최소 변경 수정본이다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "시선이 정면 카메라를 향해 크게 부릅뜬 채 고정되어 있다. 눈동자가 렌즈를 똑바로 응시하며, 아래쪽 손을 내려다보는 방향성이 전혀 없다. 샷 텍스트가 요구한 '자신의 두 손을 내려다보는' 시선이 성립하지 않는다. 두 손은 프레임 하단에서 손바닥을 카메라 쪽으로 펼쳐 얼굴 양옆에 올라와 있고, 그 자체는 화면 안에 남아 있다.",
      "built_space": "배경은 참고 스틸과 같은 고깃집 식탁: 창틀(밤 바다), 스테인리스 물통, 생고기 접시 두 개, 오른쪽 구이용 그릴, 반찬 그릇, 나무 의자 등받이. 개수와 배치가 참고 스틸의 공간과 정합하며 다른 손님은 프레임에서 제외되어 있다. 배경 요소가 과대하게 확대된 곳은 없다.",
      "entities": "20대 한국 여성, 긴 검은 머리(흐트러진 잔머리), 참고 인물과 유사한 얼굴형. 입가와 두 손이 붉은 생피로 범벅. 목걸이 없음(계약대로). 상의는 이전 스틸과 같은 베이지-회색 니트. 다만 눈이 홍채 주변 흰자가 과도하게 드러나 만화적으로 확대된 인상, 눈 크기가 부자연스럽게 커져 인물 동일성이 다소 흔들린다.",
      "hard_violations": [],
      "physics": "두 손은 팔뚝으로 지지되어 얼굴 앞에 들려 있고 손목-팔의 연결이 보인다. 상체는 식탁 앞에 앉은 자세로 무게가 실려 있다. 공중에 떠 있는 물체 없음. 피는 손가락과 손등에 젖은 광택으로 흐르고 손등 아래로 뭉침이 보여 액체로 읽힌다."
     },
     {
      "label": "B",
      "direction": "시선이 정면보다 살짝 아래로 내려가 있고 눈동자가 아래 방향(들어올린 두 손 쪽)으로 향해 있어, '자신의 피범벅 손을 내려다본다'는 샷 텍스트의 시선 타깃이 성립한다. 두 손은 프레임 하단에서 얼굴 양옆으로 올라와 상단이 잘리지 않고 유지된다.",
      "built_space": "A와 동일한 고깃집 식탁 공간: 창틀·스테인리스 물통·생고기 접시 2개·오른쪽 그릴·반찬 그릇·의자 등받이. 참고 스틸의 재질과 조명(어두운 실내, 따뜻한 저조도)과 일치. 다른 인물 없음, 배경 스케일 정상.",
      "entities": "20대 한국 여성, 긴 검은 머리, 참고 인물과 일치하는 얼굴 비례. 눈은 놀람으로 크게 떴으나 홍채·공막 비율이 해부학적으로 정상 범위이며, 입가와 두 손의 선혈, 목걸이 부재 모두 계약 충족. 의상은 이전 스틸과 동일한 니트.",
      "hard_violations": [],
      "physics": "두 손은 팔로 받쳐져 얼굴 앞에 들려 있고, 상체는 착석 자세. 무지지 부유물 없음. 피는 손가락 능선을 따라 젖은 반사와 흐름을 보이며 스티커처럼 붙어 있지 않다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "이전 스틸의 장소·조명·핏범벅 상태를 그대로 잇고, 하단에 두 손 윗부분을 남긴 초근접 얼굴 클로즈업 구도를 충족하며 눈은 자연스러운 인체 비율을 유지 — 시선이 완전한 부감이 아닌 점만 아쉽다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "A와 동일 프레임이나 계약이 요구하지 않은 눈 확대·홍채 과장과 피부 리터치가 더해져 '눈알을 키우거나 변형하지 말라'는 조항을 잠식하고, 시선의 하향감마저 약화되어 손해가 이득보다 크다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.639
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.639
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "이전 스틸의 장소·조명·핏범벅 상태를 그대로 잇고, 하단에 두 손 윗부분을 남긴 초근접 얼굴 클로즈업 구도를 충족하며 눈은 자연스러운 인체 비율을 유지 — 시선이 완전한 부감이 아닌 점만 아쉽다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "A와 동일 프레임이나 계약이 요구하지 않은 눈 확대·홍채 과장과 피부 리터치가 더해져 '눈알을 키우거나 변형하지 말라'는 조항을 잠식하고, 시선의 하향감마저 약화되어 손해가 이득보다 크다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S63sh8"
  }
 },
 "S64sh3::bgfirst_bg": {
  "input_fingerprint": "c53e1bd9cb442d1b",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 강영란(흰 머리 여자)의 주변을 에워싼 채 굳은 표정으로 어두운 숲을 향해 한쪽 발을 내디딘 mid-stride 자세의 금월도 마을 주민들의 모습.\n\nLOCATION (lock): A pitch-dark black-pine trail climbing toward the ritual clearing under a nearly full moon.\n\nTIME OF DAY (lock): night, near-full moon.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height on the path centerline, continue dollying backward in a wide frontal composition as 강영란(흰 머리 여자) advances near center and the villagers close around her in staggered depth. Catch the group with one foot committed forward, allowing the dark forest path to converge behind them while 영란’s unhesitating stride acts as the visual spearhead.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강영란 in the middle-center of the frame, midground, moves toward forest path ahead; 숲길의 진행 방향 in the lower-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 금월도 숲길 (Dark and occupied by the ascending procession); used as Creates the central movement axis and converging depth behind the advancing group; 어두운 숲 (Surrounds the path at night); used as Encloses the villagers and limits the visible escape routes around the path.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Very low-key nighttime illumination renders the procession in hard separation from the dark forest, with 영란’s red shaman robe retained as the principal color accent.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 강영란(흰 머리 여자)의 주변을 에워싼 채 굳은 표정으로 어두운 숲을 향해 한쪽 발을 내디딘 mid-stride 자세의 금월도 마을 주민들의 모습.\n\nLOCATION (lock): A pitch-dark black-pine trail climbing toward the ritual clearing under a nearly full moon.\n\nTIME OF DAY (lock): night, near-full moon.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height on the path centerline, continue dollying backward in a wide frontal composition as 강영란(흰 머리 여자) advances near center and the villagers close around her in staggered depth. Catch the group with one foot committed forward, allowing the dark forest path to converge behind them while 영란’s unhesitating stride acts as the visual spearhead.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강영란 in the middle-center of the frame, midground, moves toward forest path ahead; 숲길의 진행 방향 in the lower-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 금월도 숲길 (Dark and occupied by the ascending procession); used as Creates the central movement axis and converging depth behind the advancing group; 어두운 숲 (Surrounds the path at night); used as Encloses the villagers and limits the visible escape routes around the path.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Very low-key nighttime illumination renders the procession in hard separation from the dark forest, with 영란’s red shaman robe retained as the principal color accent.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S64sh3__bgfirst_bg.png",
  "asset_id": "7e81d567-1e25-4dba-ba76-f5ffce375701",
  "input_asset_ids": [
   "3ef9c132-4831-4bc3-bb2a-58e5b3a2e3d9",
   "26370ef0-070a-470c-beac-42f57a83eef4"
  ]
 },
 "S64sh3": {
  "input_fingerprint": "5d4c67ec192042b9",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night, near-full moon.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)의 주변을 에워싼 채 굳은 표정으로 어두운 숲을 향해 한쪽 발을 내디딘 mid-stride 자세의 금월도 마을 주민들의 모습.\n\nLOCATION (lock): A pitch-dark black-pine trail climbing toward the ritual clearing under a nearly full moon. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height on the path centerline, continue dollying backward in a wide frontal composition as 강영란(흰 머리 여자) advances near center and the villagers close around her in staggered depth. Catch the group with one foot committed forward, allowing the dark forest path to converge behind them while 영란’s unhesitating stride acts as the visual spearhead.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강영란 in the middle-center of the frame, midground, moves toward forest path ahead; 숲길의 진행 방향 in the lower-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 금월도 숲길 (Dark and occupied by the ascending procession); used as Creates the central movement axis and converging depth behind the advancing group; 어두운 숲 (Surrounds the path at night); used as Encloses the villagers and limits the visible escape routes around the path.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Very low-key nighttime illumination renders the procession in hard separation from the dark forest, with 영란’s red shaman robe retained as the principal color accent.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran continues wearing the red shaman robe as the villagers escort her through the forest.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (Korean woman in her late 60s, neatly twisted white hair, graceful dignified face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night, near-full moon.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)의 주변을 에워싼 채 굳은 표정으로 어두운 숲을 향해 한쪽 발을 내디딘 mid-stride 자세의 금월도 마을 주민들의 모습.\n\nLOCATION (lock): A pitch-dark black-pine trail climbing toward the ritual clearing under a nearly full moon. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height on the path centerline, continue dollying backward in a wide frontal composition as 강영란(흰 머리 여자) advances near center and the villagers close around her in staggered depth. Catch the group with one foot committed forward, allowing the dark forest path to converge behind them while 영란’s unhesitating stride acts as the visual spearhead.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강영란 in the middle-center of the frame, midground, moves toward forest path ahead; 숲길의 진행 방향 in the lower-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 금월도 숲길 (Dark and occupied by the ascending procession); used as Creates the central movement axis and converging depth behind the advancing group; 어두운 숲 (Surrounds the path at night); used as Encloses the villagers and limits the visible escape routes around the path.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Very low-key nighttime illumination renders the procession in hard separation from the dark forest, with 영란’s red shaman robe retained as the principal color accent.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran continues wearing the red shaman robe as the villagers escort her through the forest.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (Korean woman in her late 60s, neatly twisted white hair, graceful dignified face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night, near-full moon.\n\nSHOT TEXT (authoritative, Korean): 강영란(흰 머리 여자)의 주변을 에워싼 채 굳은 표정으로 어두운 숲을 향해 한쪽 발을 내디딘 mid-stride 자세의 금월도 마을 주민들의 모습.\n\nLOCATION (lock): A pitch-dark black-pine trail climbing toward the ritual clearing under a nearly full moon. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At chest height on the path centerline, continue dollying backward in a wide frontal composition as 강영란(흰 머리 여자) advances near center and the villagers close around her in staggered depth. Catch the group with one foot committed forward, allowing the dark forest path to converge behind them while 영란’s unhesitating stride acts as the visual spearhead.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강영란 in the middle-center of the frame, midground, moves toward forest path ahead; 숲길의 진행 방향 in the lower-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 금월도 숲길 (Dark and occupied by the ascending procession); used as Creates the central movement axis and converging depth behind the advancing group; 어두운 숲 (Surrounds the path at night); used as Encloses the villagers and limits the visible escape routes around the path.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Very low-key nighttime illumination renders the procession in hard separation from the dark forest, with 영란’s red shaman robe retained as the principal color accent.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran continues wearing the red shaman robe as the villagers escort her through the forest.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (Korean woman in her late 60s, neatly twisted white hair, graceful dignified face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S64sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S64sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:772891>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송_숲길_cf7470.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:772891>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "정면 와이드·중앙 영란·mid-stride는 잘 살렸으나 주민 의상이 사극/근대풍이라 '현대 한국' 설정과 충돌하고, 배경의 달이 실제 스케일보다 크게 전경화되어 크기 규칙을 어겼으며 다수 인물이 렌즈를 직시한다.  ★위반: [gpt] 브리프에 없는 휴대 등불 두 개와 영란의 막대기 소품이 추가되어 있다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "로케이션의 좌측 석주·바위·흑송 구조를 그대로 살린 야간 와이드 정면 구도에, 현대 한국 주민 복장과 붉은 무복의 영란이 중앙 미드그라운드에서 선두로 걷는 mid-stride를 자연스럽게 realize했다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.306,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.056,
      "B": 2.0
     },
     "violations": {
      "A": [
       "[gpt] 브리프에 없는 휴대 등불 두 개와 영란의 막대기 소품이 추가되어 있다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "정면 와이드·중앙 영란·mid-stride는 잘 살렸으나 주민 의상이 사극/근대풍이라 '현대 한국' 설정과 충돌하고, 배경의 달이 실제 스케일보다 크게 전경화되어 크기 규칙을 어겼으며 다수 인물이 렌즈를 직시한다.  ★위반: [gpt] 브리프에 없는 휴대 등불 두 개와 영란의 막대기 소품이 추가되어 있다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "로케이션의 좌측 석주·바위·흑송 구조를 그대로 살린 야간 와이드 정면 구도에, 현대 한국 주민 복장과 붉은 무복의 영란이 중앙 미드그라운드에서 선두로 걷는 mid-stride를 자연스럽게 realize했다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 길 중앙선상 가슴 높이에서 정면으로 후퇴하며, 붉은 무복의 백발 여성(강영란)이 중앙에 위치해 카메라 쪽(즉 숲길의 진행 방향)으로 전진한다. 주민들 대부분의 시선은 전방(카메라 방향)으로 고정되어 있고, 굳은 무표정으로 정면을 응시한다. 시선 타깃이 '어두운 숲길 진행 방향'과 일치한다. 다만 일부 인물의 시선이 렌즈를 정면으로 바라보는 느낌이 약간 있다.",
      "built_space": "구조물 내부 장면이 아님. 로케이션은 흑송 숲길로, 좌측 전경에 이끼 낀 선돌/바위가 레퍼런스와 동일한 배치로 서 있고, 배경은 곧게 뻗은 소나무 줄기와 마른 덤불로 채워져 있다. 뒤로 길이 수렴하는 원근이 형성되어 있으며, 배경 요소가 과장되게 확대된 곳은 없다. 사람과 나무의 실제 스케일 비례가 자연스럽다.",
      "entities": "강영란: 60대 후반 한국 여성, 단정하게 틀어 올린 백발, 품위 있는 얼굴 — 레퍼런스와 상당히 일치. 붉은 무복(붉은 두루마기형 겉옷 + 흰 치마/버선)을 착용해 CARRIED STATE 충족. 주변 주민들은 현대 한국 시골 주민 복장(패딩, 플리스, 점퍼)으로 '현대 한국' 설정과 부합. 모두 한국인으로 보이며 눈은 정상 해부학적 인간의 눈. 텍스트/워터마크 없음.",
      "hard_violations": [],
      "physics": "전원이 지면 위에 있고, 각 인물의 한쪽 발이 앞으로 내디뎌진 mid-stride 자세다. 뒷발이 땅에 붙어 체중을 지지하고 앞발이 착지하는 무게 이동이 읽히며, 팔의 스윙도 보행과 대응한다. 공중에 뜬 인물·물체 없음. 붉은 겉옷의 드레이프와 주름, 낙엽/솔잎 바닥의 질감이 물리적으로 자연스럽다. 다만 걸음 폭이 다소 얕아 stride가 약간 절제된 편."
     },
     {
      "label": "B",
      "direction": "카메라는 정면 후퇴 구도로 강영란 중앙 전진, 주민들은 전방(카메라 방향)을 향한다. 다수 인물의 시선이 렌즈를 직시하는 정도가 강해 '어두운 숲을 향한' 응시보다 카메라 정면 응시로 읽히는 얼굴이 여럿 있다. 랜턴은 손에 들려 아래로 늘어져 있어 방향성 자체는 정상.",
      "built_space": "숲길 로케이션. 좌측 이끼 선돌/바위가 레퍼런스와 유사하게 배치됨. 그러나 배경 상단 중앙에 보름달이 나무 사이로 매우 크고 선명한 원반으로 떠 있는데, 실제 스케일 대비 과도하게 크고 밝게 부각되어 배경 요소가 전경적 존재감으로 확대된 인상이다. 뒤로 수렴하는 길의 원근은 A보다 약하고 인물들이 좌우로 넓게 펼쳐진 열(列)에 가깝다.",
      "entities": "강영란: 백발을 틀어 올린 60대 후반 여성, 붉은 무복 착용 — 레퍼런스와 얼굴 유사도 준수. 그러나 주변 주민들이 전부 회색 무명 저고리·바지의 사극(조선/근대) 복장에 짚신 차림이어서, '현대 한국 금월도 마을 주민'이라는 설정과 어긋난다. 랜턴 소품도 시대 불일치를 강화한다. 텍스트/워터마크 없음.",
      "hard_violations": [],
      "physics": "인물들은 지면에 발을 딛고 있으며 대부분 한쪽 발이 앞으로 나간 보행 자세. 좌측 두 남자는 랜턴을 손에 쥐고 있어 지지가 확인된다. 다만 왼쪽 두 인물의 자세는 상체가 앞으로 과하게 숙여져 걷기보다 몸을 웅크린 어색한 포즈이며, 중앙 인물의 손에 들린 짧은 막대(부채/막대 소품)는 파지가 모호하게 렌더링되어 있다. 공중 부양이나 지지 없는 물체는 없다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "가슴 높이 정면 후퇴 와이드, 중앙 중경의 붉은 무복 강영란과 엇갈린 깊이로 에워싼 주민들, 전원 한쪽 발을 내디딘 mid-stride가 모두 충족되고 현대 한국 마을 주민 복장·로케이션 선돌까지 레퍼런스와 일치한다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "구도와 mid-stride는 유사하나 주민들이 전원 사극풍 무명옷·짚신·랜턴 차림이라 현대 한국 설정과 어긋나고, 배경 보름달이 실제 스케일 이상으로 크게 부각되며 렌즈 직시 얼굴이 많아 '어두운 숲을 향한' 응시가 약하다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "가슴 높이 정면 후퇴 와이드, 중앙 중경의 붉은 무복 강영란과 엇갈린 깊이로 에워싼 주민들, 전원 한쪽 발을 내디딘 mid-stride가 모두 충족되고 현대 한국 마을 주민 복장·로케이션 선돌까지 레퍼런스와 일치한다."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "구도와 mid-stride는 유사하나 주민들이 전원 사극풍 무명옷·짚신·랜턴 차림이라 현대 한국 설정과 어긋나고, 배경 보름달이 실제 스케일 이상으로 크게 부각되며 렌즈 직시 얼굴이 많아 '어두운 숲을 향한' 응시가 약하다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "카메라는 길 중앙선상 가슴 높이에서 정면으로 후퇴하며, 붉은 무복의 백발 여성(강영란)이 중앙에 위치해 카메라 쪽(즉 숲길의 진행 방향)으로 전진한다. 주민들 대부분의 시선은 전방(카메라 방향)으로 고정되어 있고, 굳은 무표정으로 정면을 응시한다. 시선 타깃이 '어두운 숲길 진행 방향'과 일치한다. 다만 일부 인물의 시선이 렌즈를 정면으로 바라보는 느낌이 약간 있다.",
      "built_space": "구조물 내부 장면이 아님. 로케이션은 흑송 숲길로, 좌측 전경에 이끼 낀 선돌/바위가 레퍼런스와 동일한 배치로 서 있고, 배경은 곧게 뻗은 소나무 줄기와 마른 덤불로 채워져 있다. 뒤로 길이 수렴하는 원근이 형성되어 있으며, 배경 요소가 과장되게 확대된 곳은 없다. 사람과 나무의 실제 스케일 비례가 자연스럽다.",
      "entities": "강영란: 60대 후반 한국 여성, 단정하게 틀어 올린 백발, 품위 있는 얼굴 — 레퍼런스와 상당히 일치. 붉은 무복(붉은 두루마기형 겉옷 + 흰 치마/버선)을 착용해 CARRIED STATE 충족. 주변 주민들은 현대 한국 시골 주민 복장(패딩, 플리스, 점퍼)으로 '현대 한국' 설정과 부합. 모두 한국인으로 보이며 눈은 정상 해부학적 인간의 눈. 텍스트/워터마크 없음.",
      "hard_violations": [],
      "physics": "전원이 지면 위에 있고, 각 인물의 한쪽 발이 앞으로 내디뎌진 mid-stride 자세다. 뒷발이 땅에 붙어 체중을 지지하고 앞발이 착지하는 무게 이동이 읽히며, 팔의 스윙도 보행과 대응한다. 공중에 뜬 인물·물체 없음. 붉은 겉옷의 드레이프와 주름, 낙엽/솔잎 바닥의 질감이 물리적으로 자연스럽다. 다만 걸음 폭이 다소 얕아 stride가 약간 절제된 편."
     },
     {
      "label": "A",
      "direction": "카메라는 정면 후퇴 구도로 강영란 중앙 전진, 주민들은 전방(카메라 방향)을 향한다. 다수 인물의 시선이 렌즈를 직시하는 정도가 강해 '어두운 숲을 향한' 응시보다 카메라 정면 응시로 읽히는 얼굴이 여럿 있다. 랜턴은 손에 들려 아래로 늘어져 있어 방향성 자체는 정상.",
      "built_space": "숲길 로케이션. 좌측 이끼 선돌/바위가 레퍼런스와 유사하게 배치됨. 그러나 배경 상단 중앙에 보름달이 나무 사이로 매우 크고 선명한 원반으로 떠 있는데, 실제 스케일 대비 과도하게 크고 밝게 부각되어 배경 요소가 전경적 존재감으로 확대된 인상이다. 뒤로 수렴하는 길의 원근은 A보다 약하고 인물들이 좌우로 넓게 펼쳐진 열(列)에 가깝다.",
      "entities": "강영란: 백발을 틀어 올린 60대 후반 여성, 붉은 무복 착용 — 레퍼런스와 얼굴 유사도 준수. 그러나 주변 주민들이 전부 회색 무명 저고리·바지의 사극(조선/근대) 복장에 짚신 차림이어서, '현대 한국 금월도 마을 주민'이라는 설정과 어긋난다. 랜턴 소품도 시대 불일치를 강화한다. 텍스트/워터마크 없음.",
      "hard_violations": [],
      "physics": "인물들은 지면에 발을 딛고 있으며 대부분 한쪽 발이 앞으로 나간 보행 자세. 좌측 두 남자는 랜턴을 손에 쥐고 있어 지지가 확인된다. 다만 왼쪽 두 인물의 자세는 상체가 앞으로 과하게 숙여져 걷기보다 몸을 웅크린 어색한 포즈이며, 중앙 인물의 손에 들린 짧은 막대(부채/막대 소품)는 파지가 모호하게 렌더링되어 있다. 공중 부양이나 지지 없는 물체는 없다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 11,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "정면 와이드·중앙 영란·mid-stride는 잘 살렸으나 주민 의상이 사극/근대풍이라 '현대 한국' 설정과 충돌하고, 배경의 달이 실제 스케일보다 크게 전경화되어 크기 규칙을 어겼으며 다수 인물이 렌즈를 직시한다.  ★위반: [gpt] 브리프에 없는 휴대 등불 두 개와 영란의 막대기 소품이 추가되어 있다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "로케이션의 좌측 석주·바위·흑송 구조를 그대로 살린 야간 와이드 정면 구도에, 현대 한국 주민 복장과 붉은 무복의 영란이 중앙 미드그라운드에서 선두로 걷는 mid-stride를 자연스럽게 realize했다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_해송_숲길_cf7470.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:772891>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "강영란이 두 발을 가지런히 모으고 팔을 늘어뜨린 채 서 있어, 지시된 힘찬 걸음(mid-stride) 자세가 아닙니다.",
     "fix_en": "Redraw Youngran's lower body, red robe hem, white skirt, and feet to show her in mid-stride, with one foot confidently stepping forward toward the camera. Preserve her face, upper body, the surrounding villagers and their poses, the dark forest setting, the lighting, and the exact wide-shot camera framing.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw Youngran's lower body, red robe hem, white skirt, and feet to show her in mid-stride, with one foot confidently stepping forward toward the camera. Preserve her face, upper body, the surrounding villagers and their poses, the dark forest setting, the lighting, and the exact wide-shot camera framing.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "와이드 정면 후진 달리, 중앙 중경의 강영란과 이를 에워싼 mid-stride 주민들, 붉은 무복 유일 색 악센트, 로케이션 선돌·흑송 일치 — 계약 항목을 군더더기 없이 충족하고 의상 드레이프도 물리적으로 자연스럽다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "구도·방향·인물 구성은 A와 동일하나, 계약이 요구하지 않은 붉은 무복의 대형 망토식 확장이 추가되어 캐릭터 레퍼런스의 실루엣에서 벗어나고 천이 판지처럼 뻣뻣해 MATERIAL REALISM에서 손해를 본다 — 얻은 것 없이 바꾼 변경이므로 감점."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.889,
      "B": 1.75
     },
     "adjusted": {
      "A": 1.889,
      "B": 1.75
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "B"
     },
     "agreed": false
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "와이드 정면 후진 달리, 중앙 중경의 강영란과 이를 에워싼 mid-stride 주민들, 붉은 무복 유일 색 악센트, 로케이션 선돌·흑송 일치 — 계약 항목을 군더더기 없이 충족하고 의상 드레이프도 물리적으로 자연스럽다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "구도·방향·인물 구성은 A와 동일하나, 계약이 요구하지 않은 붉은 무복의 대형 망토식 확장이 추가되어 캐릭터 레퍼런스의 실루엣에서 벗어나고 천이 판지처럼 뻣뻣해 MATERIAL REALISM에서 손해를 본다 — 얻은 것 없이 바꾼 변경이므로 감점."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·로케이션·야간 조도는 충족하나, 붉은 무복이 케이프처럼 좌우로 크게 펼쳐져 천의 무게·드레이프가 어긋나고 손이 사라져 '붉은 무복을 계속 입고 이동한다'는 CARRIED STATE와 물질감 조항에서 손해를 본다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일 인물·동일 배치·동일 야간 저조도를 유지하면서 붉은 무복을 실제 두루마기처럼 몸에 자연스럽게 드리워 재질·중량감과 CARRIED STATE를 정확히 지켰고, mid-stride 도열과 수렴하는 숲길 축도 그대로 충족한다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.75,
      "B": 1.889
     },
     "adjusted": {
      "A": 1.75,
      "B": 1.889
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "A"
     },
     "agreed": false
    }
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "프레이밍·로케이션·야간 조도는 충족하나, 붉은 무복이 케이프처럼 좌우로 크게 펼쳐져 천의 무게·드레이프가 어긋나고 손이 사라져 '붉은 무복을 계속 입고 이동한다'는 CARRIED STATE와 물질감 조항에서 손해를 본다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "동일 인물·동일 배치·동일 야간 저조도를 유지하면서 붉은 무복을 실제 두루마기처럼 몸에 자연스럽게 드리워 재질·중량감과 CARRIED STATE를 정확히 지켰고, mid-stride 도열과 수렴하는 숲길 축도 그대로 충족한다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 12
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S64sh3__bgfirst_bg.png",
   "bg_asset_id": "7e81d567-1e25-4dba-ba76-f5ffce375701",
   "bg_record_key": "S64sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "해송 숲길",
   "groupbg_asset_id": "26370ef0-070a-470c-beac-42f57a83eef4"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S70sh4": {
  "input_fingerprint": "045d630275e9806d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어둠 속을 향해 앞발을 내딛은 자세의 김지후(강수리영) 뒷모습을 보며 황당함에 입을 떡 벌린 채 굳어있는 정인우의 얼굴 클로즈업.\n\nLOCATION (lock): The exterior cave threshold where the forest path meets the pitch-black opening. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: End the dolly-in at 정인우's face height from a front three-quarter angle, tightening until his stunned face occupies most of the center-left frame without exceeding a natural close-up. Keep 김지후(강수리영)'s back and forward-stepping leg along the adjacent right sightline, aimed into the cave, so his reaction and her decision remain in one spatial axis.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 정인우 in the middle-left of the frame, foreground, looks toward 김지후(강수리영) entering the cave; 김지후(강수리영) in the middle-right of the frame, midground, moves toward cave interior.\n- KEY BACKGROUND ELEMENTS: 자연동굴 입구 (어둠 속에 열려 있음) — The opening is seen obliquely past 김지후(강수리영), with the interior receding away from camera; used as Dark destination aligned behind 김지후(강수리영)'s receding back; 덧대어진 나무 기둥 (동굴 입구에 덧대어져 있음) — Only oblique side faces remain visible near the cave edge; used as Partial edge structure retaining continuity with the preceding wide shot.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key night contrast keeps 정인우's reaction readable while 김지후(강수리영) falls toward the cave's deeper darkness.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same supported cave mouth, darkness, moonlit exterior, and the woman's step into the entrance. Exclude a broad rear view of both figures; focus on the man's stunned reaction behind her.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu enters without her necklace; Inwoo follows with his injured arm still casted and his backpack and flashlight retained.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "C": 6,
   "B": 5
  },
  "selected": "A",
  "ranking": [
   "A",
   "C",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "유일하게 '클로즈업'과 '중앙-좌측 전경 얼굴' 지시를 충족하고, 놀란 표정과 우측으로 멀어지는 지후의 시선축이 한 공간축에 놓여 동굴 어둠까지 이어진다; 시선이 카메라를 살짝 스치는 점만 아쉽다."
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "장소·의상·깁스·손전등 등 지속 요소는 잘 지켰으나 프레임이 미디엄에 가까워 클로즈업 지시를 어기고, 시선이 거의 정면 카메라를 향해 지후를 보는 축이 성립하지 않는다."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "가장 헐거운 프레이밍(미디엄 숏)으로 클로즈업 지시에서 가장 멀고, 시선이 좌측 프레임 밖을 향해 지후를 바라보는 반응축이 끊긴다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S70sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:787232>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 와이드 샷 래퍼런스에서는 남자의 가슴 쪽에 아무런 끈이 없었으나, 현재 샷에서는 가슴을 가로지르는 굵은 흰색 끈(슬링)이 추가되어 연속성에 어긋납니다.",
     "fix_en": "Remove the thick white cloth strap crossing diagonally over the man's chest. Fill that space with the continuous plain olive-green fabric of his jacket. Preserve the man's face, his expression, the rest of his jacket, his dark backpack straps, the woman in the background, the cave entrance, and the exact current lighting and framing.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the thick white cloth strap crossing diagonally over the man's chest. Fill that space with the continuous plain olive-green fabric of his jacket. Preserve the man's face, his expression, the rest of his jacket, his dark backpack straps, the woman in the background, the cave entrance, and the exact current lighting and framing.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 구도·로케이션 연속성을 지키면서 흰 깁스와 배낭 스트랩까지 보여 CARRIED STATE를 충족하고, 김지후의 앞발 내딛음도 명확하다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "구도는 A와 거의 같으나 계약이 요구한 깁스한 팔이 사라지고 김지후의 '앞발 내딛은 자세'도 약해져, 요청되지 않은 손실만 발생했다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.889,
      "B": 1.75
     },
     "adjusted": {
      "A": 1.889,
      "B": 1.75
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "B"
     },
     "agreed": false
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 구도·로케이션 연속성을 지키면서 흰 깁스와 배낭 스트랩까지 보여 CARRIED STATE를 충족하고, 김지후의 앞발 내딛음도 명확하다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "구도는 A와 거의 같으나 계약이 요구한 깁스한 팔이 사라지고 김지후의 '앞발 내딛은 자세'도 약해져, 요청되지 않은 손실만 발생했다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 9,
      "verdict_ko": "정확한 얼굴 클로즈업, 지후를 향한 시선과 동굴로 이어지는 보행축을 유지하면서 프레임 밖 깁스를 억지로 표시하지 않아, 불필요한 변경이 없는 A가 가장 충실하다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "핵심 구도와 방향은 A와 같고 부상을 암시하는 흰 슬링을 얻었지만, 이전 장면에 없던 목걸이형 장비를 의상 위에 추가한 비용이 있어 변경이 적은 A보다 낮다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.778
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.778
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "정확한 얼굴 클로즈업, 지후를 향한 시선과 동굴로 이어지는 보행축을 유지하면서 프레임 밖 깁스를 억지로 표시하지 않아, 불필요한 변경이 없는 A가 가장 충실하다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "핵심 구도와 방향은 A와 같고 부상을 암시하는 흰 슬링을 얻었지만, 이전 장면에 없던 목걸이형 장비를 의상 위에 추가한 비용이 있어 변경이 적은 A보다 낮다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 15
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S70sh1"
  }
 },
 "S71sh19": {
  "input_fingerprint": "8b7390568eb9e205",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 오열하는 강민숙(홍은미)을 바라보며 충격에 빠져 할 말을 잃고 굳어버린 김지후(강수리영)와 정인우의 전신.\n\nLOCATION (lock): The blood-stained inner cave chamber around the chained holding recess. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Cut to a static reverse wide frame at chest height from near and slightly behind 강민숙(홍은미), placing her hunched back low in the foreground while 김지후(강수리영) and 정인우 remain visible head to foot several steps away. Keep the two shocked figures separated across the middle ground, their attention converging toward the sobbing captive rather than toward the lens.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 강민숙(홍은미) in the lower-center of the frame, foreground; 김지후(강수리영) in the middle-left of the frame, midground, looks toward sobbing 강민숙(홍은미); 정인우 in the middle-right of the frame, midground, looks toward sobbing 강민숙(홍은미).\n- KEY BACKGROUND ELEMENTS: 동굴 바닥 (인골과 아직 마르지 않은 피웅덩이가 놓여 있음); used as Floor plane maintaining the physical distance between captive and rescuers; 쇠사슬 (강민숙(홍은미)을 묶고 있음) — The chain crosses the foreground around 강민숙(홍은미)'s body and leads toward the cave corner; used as Foreground restraint partially visible around the sobbing captive; 동굴 내부 (깊은 어둠으로 이어져 있음); used as Dark enclosure behind the two silent figures.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Flashlight illumination separates the two frozen figures and the captive's foreground outline from the otherwise dark cave.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same cave corner, chains, flashlight illumination, bones, and captive woman's position. Exclude an isolated view of the captive; show the two rescuers frozen while she sobs.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk remains chained and unable to leave; Jihu’s necklace remains absent, and Inwoo’s arm remains casted.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face); 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 6,
   "B": 8,
   "C": 3
  },
  "selected": "B",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "리버스 와이드 구도와 두 인물 전신·접지는 충족하지만, 손전등을 렌즈 정면으로 겨눠 광원 방향이 오열하는 강민숙에게 떨어지지 않고 김지후의 캡 모자 누락 등 레퍼런스 일치도가 B보다 떨어진다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "강민숙의 굽은 등을 하단 전경에 두고 좌·우로 분리된 두 인물이 전신으로 그녀를 내려다보는 리버스 와이드를 정확히 구현했고, 캡·캐스트·목걸이 부재 등 의상/캐리 계약과 쇠사슬·인골·조명의 장소 락도 가장 충실하다."
   },
   {
    "label": "C",
    "score": 3,
    "verdict_ko": "전경 인물의 상체가 해부학적으로 불가능한 평면 덩어리로 붕괴하고 쇠사슬이 지지점 없이 공중에 떠 전경을 과대 점유하며, 김지후 목걸이까지 등장해 '목걸이 부재' 계약을 위반했다.  ★위반: [claude-opus] 전경 인물(강민숙)의 상체·머리 연결이 해부학적으로 불가능한 평면 덩어리로 렌더링됨 / [claude-opus] 전경 쇠사슬이 지지점 없이 공중에 떠 프레임을 가로지르며 실제 크기를 넘어 전경 오브젝트로 확대됨"
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S71sh14_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:813104>"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:787232>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 샷(레퍼런스)에서 강민숙은 벽에 기대어 바닥에 앉아 있었으나, 현재 샷에서는 동굴 한가운데 서서 허리를 굽히고 있어 고정되어야 할 자세와 위치가 일치하지 않습니다.",
     "fix_en": "Redraw Kang Minsuk's body in the foreground to appear as a hunched, seated mass resting on the floor, keeping her bloody sweater, while strictly preserving the two standing characters, the cave background, and the current camera framing.",
     "needs_regeneration": true
    },
    {
     "issue_ko": "화면 왼쪽 김지후의 오른쪽 소매 끝에 손이 그려지지 않아 팔이 절단된 것처럼 보입니다.",
     "fix_en": "Redraw Kim Jihu's right wrist to show a natural human hand resting at her side, while leaving her face, clothing, the other characters, and the cave entirely unchanged."
    },
    {
     "issue_ko": "화면 앞쪽의 쇠사슬이 묶인 금속 목줄로 이어지지 않고 강민숙의 머리카락과 뒤통수에 물리적으로 융합되어 있습니다.",
     "fix_en": "Redraw the top end of the chain so it connects logically to a thick, rusted metal collar around the captive's neck, ensuring it does not fuse with her hair, while keeping the rest of the image exactly as it is."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw Kang Minsuk's body in the foreground to appear as a hunched, seated mass resting on the floor, keeping her bloody sweater, while strictly preserving the two standing characters, the cave background, and the current camera framing.\n- Redraw Kim Jihu's right wrist to show a natural human hand resting at her side, while leaving her face, clothing, the other characters, and the cave entirely unchanged.\n- Redraw the top end of the chain so it connects logically to a thick, rusted metal collar around the captive's neck, ensuring it does not fuse with her hair, while keeping the rest of the image exactly as it is.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 중간 좌측, 정인우는 중간 우측에 서서 두 사람 모두 시선이 프레임 하단 중앙의 웅크린 강민숙 쪽(약간 아래·안쪽)으로 향한다. 렌즈를 정면으로 응시하지 않고 시선이 전경의 포로에게 수렴한다. 카메라는 강민숙의 등 뒤 약간 위·가슴 높이에서 정면으로 두 사람을 잡는 리버스 와이드로 읽힌다.",
      "built_space": "동굴 통로형 공간: 좌우 젖은 암벽, 중앙 안쪽은 깊은 어둠, 바닥엔 자갈·인골·아직 마르지 않은 검붉은 피웅덩이. 좌측 하단에서 쇠사슬이 들어와 전경 강민숙의 몸(등·목 쪽)을 지나 동굴 구석으로 이어진다. 사슬 고리는 하나의 연속선으로 읽히며 중복된 구조물은 없다. 이전 스틸의 재질(젖은 바위, 녹슨 사슬, 흰 뼈)이 유지된다.",
      "entities": "김지후: 20대 초 한국 여성, 긴 검은 생머리, 회색 캡, 갈색 재킷+올리브 헨리넥+짙은 카고팬츠+갈색 부츠 — 캐릭터 레퍼런스와 일치, 목걸이 없음(계약대로). 정인우: 20대 중반 한국 남성, 짙은 머리, 녹색 코트+버건디 터틀넥+회색 팬츠+갈색 구두, 왼팔에 흰 캐스트와 슬링 — 계약대로. 전경 인물은 흰 니트, 백발 섞인 헝클어진 머리, 피에 젖은 옷의 노년 여성 포로로 이전 스틸과 동일 인물로 읽힌다. 추가 인물 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 두 발이 동굴 바닥에 닿아 체중이 실려 있고, 굳은 채 팔을 늘어뜨린 정지 자세. 강민숙은 무릎을 꿇고 상체를 앞으로 숙인 자세로 바닥에 접지되어 있으며 머리·목·등의 연결이 해부학적으로 자연스럽다. 쇠사슬은 바닥에 늘어져 무게를 갖고 놓여 있다. 공중에 뜬 것 없음."
     },
     {
      "label": "B",
      "direction": "두 인물의 배치와 시선은 A와 동일하게 전경의 포로 쪽을 향한다. 다만 전경 인물의 상체가 더 낮게 눌려 있어 머리 방향과 시선 수렴점이 A보다 약간 모호하다.",
      "built_space": "동굴 구조·사슬·뼈·피웅덩이 배치는 A와 사실상 동일. 다만 중앙 안쪽 어둠이 더 넓고 좌우 암벽의 디테일이 다소 줄었으며, 전경 포로의 위치가 프레임 아래에서 조금 더 위로 올라와 인물이 화면 중앙을 크게 차지한다.",
      "entities": "김지후·정인우의 의상·머리·캐스트·목걸이 부재는 A와 동일하게 일치. 전경 포로는 흰 니트·백발이지만 목에 굵은 가죽 목줄이 새로 강조되고, 몸통이 비정상적으로 둥글게 부풀어 머리가 어깨 사이로 함몰된 듯 보여 인체 비례가 뭉개진다.",
      "hard_violations": [
       "전경 강민숙의 해부가 붕괴 — 목과 어깨 경계가 사라져 머리가 등에 파묻힌 덩어리로 읽히고 상체가 자루 형태로 뭉개져, '인간 형상을 최대한 렌더링' 조항과 물리적으로 불가능한 해부 금지에 저촉"
      ],
      "physics": "두 서 있는 인물은 정상적으로 바닥에 접지. 그러나 전경 포로는 무릎 꿇은 다리·팔의 지지 구조가 보이지 않고 둥근 몸통 덩어리가 바닥에 얹힌 형태로만 읽혀, 오열하는 인체의 지지점(무릎·손)이 확인되지 않는다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "리버스 와이드·전경 강민숙 하단 중앙·좌우 두 인물 전신과 시선 수렴, 캐스트·목걸이 부재 등 계약을 모두 충족하고 전경 포로의 인체와 사슬이 해부학·물리적으로 자연스럽게 읽힌다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "배치·의상은 A와 같으나 전경 강민숙의 목·어깨가 사라진 자루 형태 덩어리로 뭉개져 인체 형상이 붕괴됐고, 없던 가죽 목줄 강조·전경 인물 확대 등 계약이 요구하지 않은 변경이 더해져 손해가 크다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "리버스 와이드·전경 강민숙 하단 중앙·좌우 두 인물 전신과 시선 수렴, 캐스트·목걸이 부재 등 계약을 모두 충족하고 전경 포로의 인체와 사슬이 해부학·물리적으로 자연스럽게 읽힌다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "배치·의상은 A와 같으나 전경 강민숙의 목·어깨가 사라진 자루 형태 덩어리로 뭉개져 인체 형상이 붕괴됐고, 없던 가죽 목줄 강조·전경 인물 확대 등 계약이 요구하지 않은 변경이 더해져 손해가 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "김지후는 중간 좌측, 정인우는 중간 우측에 서서 두 사람 모두 시선이 프레임 하단 중앙의 웅크린 강민숙 쪽(약간 아래·안쪽)으로 향한다. 렌즈를 정면으로 응시하지 않고 시선이 전경의 포로에게 수렴한다. 카메라는 강민숙의 등 뒤 약간 위·가슴 높이에서 정면으로 두 사람을 잡는 리버스 와이드로 읽힌다.",
      "built_space": "동굴 통로형 공간: 좌우 젖은 암벽, 중앙 안쪽은 깊은 어둠, 바닥엔 자갈·인골·아직 마르지 않은 검붉은 피웅덩이. 좌측 하단에서 쇠사슬이 들어와 전경 강민숙의 몸(등·목 쪽)을 지나 동굴 구석으로 이어진다. 사슬 고리는 하나의 연속선으로 읽히며 중복된 구조물은 없다. 이전 스틸의 재질(젖은 바위, 녹슨 사슬, 흰 뼈)이 유지된다.",
      "entities": "김지후: 20대 초 한국 여성, 긴 검은 생머리, 회색 캡, 갈색 재킷+올리브 헨리넥+짙은 카고팬츠+갈색 부츠 — 캐릭터 레퍼런스와 일치, 목걸이 없음(계약대로). 정인우: 20대 중반 한국 남성, 짙은 머리, 녹색 코트+버건디 터틀넥+회색 팬츠+갈색 구두, 왼팔에 흰 캐스트와 슬링 — 계약대로. 전경 인물은 흰 니트, 백발 섞인 헝클어진 머리, 피에 젖은 옷의 노년 여성 포로로 이전 스틸과 동일 인물로 읽힌다. 추가 인물 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 두 발이 동굴 바닥에 닿아 체중이 실려 있고, 굳은 채 팔을 늘어뜨린 정지 자세. 강민숙은 무릎을 꿇고 상체를 앞으로 숙인 자세로 바닥에 접지되어 있으며 머리·목·등의 연결이 해부학적으로 자연스럽다. 쇠사슬은 바닥에 늘어져 무게를 갖고 놓여 있다. 공중에 뜬 것 없음."
     },
     {
      "label": "B",
      "direction": "두 인물의 배치와 시선은 A와 동일하게 전경의 포로 쪽을 향한다. 다만 전경 인물의 상체가 더 낮게 눌려 있어 머리 방향과 시선 수렴점이 A보다 약간 모호하다.",
      "built_space": "동굴 구조·사슬·뼈·피웅덩이 배치는 A와 사실상 동일. 다만 중앙 안쪽 어둠이 더 넓고 좌우 암벽의 디테일이 다소 줄었으며, 전경 포로의 위치가 프레임 아래에서 조금 더 위로 올라와 인물이 화면 중앙을 크게 차지한다.",
      "entities": "김지후·정인우의 의상·머리·캐스트·목걸이 부재는 A와 동일하게 일치. 전경 포로는 흰 니트·백발이지만 목에 굵은 가죽 목줄이 새로 강조되고, 몸통이 비정상적으로 둥글게 부풀어 머리가 어깨 사이로 함몰된 듯 보여 인체 비례가 뭉개진다.",
      "hard_violations": [
       "전경 강민숙의 해부가 붕괴 — 목과 어깨 경계가 사라져 머리가 등에 파묻힌 덩어리로 읽히고 상체가 자루 형태로 뭉개져, '인간 형상을 최대한 렌더링' 조항과 물리적으로 불가능한 해부 금지에 저촉"
      ],
      "physics": "두 서 있는 인물은 정상적으로 바닥에 접지. 그러나 전경 포로는 무릎 꿇은 다리·팔의 지지 구조가 보이지 않고 둥근 몸통 덩어리가 바닥에 얹힌 형태로만 읽혀, 오열하는 인체의 지지점(무릎·손)이 확인되지 않는다."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "카메라는 전경 강민숙의 등 뒤·약간 뒤쪽에서 동굴 안쪽을 향한다. 좌측 중경의 김지후는 고개를 살짝 숙여 전경의 강민숙 쪽(화면 중앙 하단)을 내려다보고, 우측의 정인우도 시선이 중앙 아래 오열하는 여자 쪽으로 향한다. 두 사람 모두 렌즈를 직시하지 않아 시선 수렴 지시는 충족. 쇠사슬은 좌측 전경에서 강민숙의 목/어깨를 지나 좌측 동굴 구석으로 이어진다.",
      "built_space": "동굴 내부: 좌우 암벽, 중앙 깊은 어둠, 바닥에 인골 다수와 자갈, 우측 하단에 아직 마르지 않은 피웅덩이 1개. 참조 스틸의 소재·질감·조명 분위기와 대체로 일치. 다만 강민숙의 몸이 전경에서 비정상적으로 크고 뭉툭하게 부풀어 있어 상체가 하나의 덩어리처럼 읽히고, 머리·목·어깨·엉덩이의 해부학적 연결이 불명확하다(등이 거대한 자루처럼 보임).",
      "entities": "김지후: 회색 캡, 갈색 워크재킷, 올리브 헨리넥, 어두운 카고팬츠, 부츠 — 캐릭터 참조 일치, 목걸이 없음(계약 충족). 정인우: 짙은 녹색 코트, 버건디 터틀넥, 브로치, 회색 슬랙스, 갈색 구두, 오른팔 캐스트+슬링 — 참조 및 캐리드 스테이트 일치. 강민숙: 뒷모습, 흰 회색 머리, 더러운 베이지 상의, 쇠사슬 — 이전 스틸과 옷/사슬 일치하나 머리가 은백색으로 더 밝고 상체 실루엣이 원본보다 훨씬 비대하다.",
      "hard_violations": [
       "전경 강민숙의 신체가 해부학적으로 불가능한 덩어리 형태 — 목/어깨/등/골반이 하나의 부푼 자루로 뭉개져 인체 구조가 성립하지 않음(사람을 형체 덩어리로 환원하지 말라는 지시 위반)"
      ],
      "physics": "김지후·정인우는 두 발이 자갈·인골 바닥에 확실히 닿아 서 있고 무게중심이 안정적. 강민숙은 바닥에 웅크려 앉아 있으나 하반신·팔·손이 전혀 보이지 않아 무엇이 그 거대한 상체를 지지하는지 읽히지 않으며, 형태가 옷 무더기처럼 부유하는 인상을 준다. 사슬은 바닥과 몸을 따라 늘어져 중력에 맞게 처짐."
     },
     {
      "label": "B",
      "direction": "동일한 리버스 와이드 축. 김지후(좌 중경)는 고개를 아래로 기울여 전경 중앙의 오열하는 강민숙을 응시하고, 정인우(우 중경)도 시선이 아래 중앙의 강민숙으로 수렴한다. 둘 다 렌즈를 보지 않음. 쇠사슬은 좌측 전경 바닥에서 뻗어 강민숙의 등을 타고 목덜미로 이어져 동굴 좌측 구석으로 향한다.",
      "built_space": "좌우 암벽과 중앙 깊은 어둠, 자갈·인골이 깔린 바닥, 우측 하단의 젖은 피웅덩이 — 이전 스틸의 장소·재질·손전등 조명 분위기 유지. 강민숙은 전경 하단 중앙에서 웅크려 등을 보이며, 등·허리·엉덩이·다리(흙 묻은 바지)까지 인체 비율이 자연스럽게 연결됨.",
      "entities": "김지후: 캡, 갈색 재킷, 올리브 헨리넥, 카고팬츠, 부츠, 목걸이 없음 — 참조 일치. 정인우: 녹색 코트, 버건디 터틀넥, 브로치, 우측 팔 캐스트와 흰 슬링 — 참조 및 캐리드 스테이트 일치. 강민숙: 뒷모습, 헝클어진 회백색 머리, 피 묻은 낡은 니트, 목·등을 감은 녹슨 사슬, 하단에 더러운 바지 — 이전 스틸의 의상·사슬 상태와 일치.",
      "hard_violations": [],
      "physics": "두 구조자는 바닥에 두 발을 딛고 정지해 서 있으며 무게 이동이 자연스럽다. 강민숙은 무릎을 굽혀 바닥에 웅크린 자세로, 상체가 앞으로 숙여지고 하반신이 바닥에 닿아 지지된다 — 오열하며 몸을 접은 동작으로 설명 가능. 사슬은 바닥의 자갈 위로 늘어져 무게에 따라 처짐."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "카메라·시선·의상·캐스트 등 계약 요소는 갖췄으나, 전경 강민숙의 상체가 목·어깨·골반 구분 없는 거대한 자루형 덩어리로 뭉개져 인체가 성립하지 않고 지지도 읽히지 않는다 — '사람을 형체 덩어리로 환원 금지' 및 해부학적 성립성 위반이 결정적."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 리버스 와이드 프레임·두 인물의 전신·오열하는 포로로의 시선 수렴을 지키면서, 전경 강민숙의 등·허리·다리가 자연스럽게 연결된 웅크린 자세로 바닥에 지지되고 사슬·인골·피웅덩이·손전등 조명 등 장소 락도 유지 — 변경은 전경 인물 형태 정상화에 국한되어 얻는 것이 잃는 것보다 크다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "카메라·시선·의상·캐스트 등 계약 요소는 갖췄으나, 전경 강민숙의 상체가 목·어깨·골반 구분 없는 거대한 자루형 덩어리로 뭉개져 인체가 성립하지 않고 지지도 읽히지 않는다 — '사람을 형체 덩어리로 환원 금지' 및 해부학적 성립성 위반이 결정적."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "동일한 리버스 와이드 프레임·두 인물의 전신·오열하는 포로로의 시선 수렴을 지키면서, 전경 강민숙의 등·허리·다리가 자연스럽게 연결된 웅크린 자세로 바닥에 지지되고 사슬·인골·피웅덩이·손전등 조명 등 장소 락도 유지 — 변경은 전경 인물 형태 정상화에 국한되어 얻는 것이 잃는 것보다 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "카메라는 전경 강민숙의 등 뒤·약간 뒤쪽에서 동굴 안쪽을 향한다. 좌측 중경의 김지후는 고개를 살짝 숙여 전경의 강민숙 쪽(화면 중앙 하단)을 내려다보고, 우측의 정인우도 시선이 중앙 아래 오열하는 여자 쪽으로 향한다. 두 사람 모두 렌즈를 직시하지 않아 시선 수렴 지시는 충족. 쇠사슬은 좌측 전경에서 강민숙의 목/어깨를 지나 좌측 동굴 구석으로 이어진다.",
      "built_space": "동굴 내부: 좌우 암벽, 중앙 깊은 어둠, 바닥에 인골 다수와 자갈, 우측 하단에 아직 마르지 않은 피웅덩이 1개. 참조 스틸의 소재·질감·조명 분위기와 대체로 일치. 다만 강민숙의 몸이 전경에서 비정상적으로 크고 뭉툭하게 부풀어 있어 상체가 하나의 덩어리처럼 읽히고, 머리·목·어깨·엉덩이의 해부학적 연결이 불명확하다(등이 거대한 자루처럼 보임).",
      "entities": "김지후: 회색 캡, 갈색 워크재킷, 올리브 헨리넥, 어두운 카고팬츠, 부츠 — 캐릭터 참조 일치, 목걸이 없음(계약 충족). 정인우: 짙은 녹색 코트, 버건디 터틀넥, 브로치, 회색 슬랙스, 갈색 구두, 오른팔 캐스트+슬링 — 참조 및 캐리드 스테이트 일치. 강민숙: 뒷모습, 흰 회색 머리, 더러운 베이지 상의, 쇠사슬 — 이전 스틸과 옷/사슬 일치하나 머리가 은백색으로 더 밝고 상체 실루엣이 원본보다 훨씬 비대하다.",
      "hard_violations": [
       "전경 강민숙의 신체가 해부학적으로 불가능한 덩어리 형태 — 목/어깨/등/골반이 하나의 부푼 자루로 뭉개져 인체 구조가 성립하지 않음(사람을 형체 덩어리로 환원하지 말라는 지시 위반)"
      ],
      "physics": "김지후·정인우는 두 발이 자갈·인골 바닥에 확실히 닿아 서 있고 무게중심이 안정적. 강민숙은 바닥에 웅크려 앉아 있으나 하반신·팔·손이 전혀 보이지 않아 무엇이 그 거대한 상체를 지지하는지 읽히지 않으며, 형태가 옷 무더기처럼 부유하는 인상을 준다. 사슬은 바닥과 몸을 따라 늘어져 중력에 맞게 처짐."
     },
     {
      "label": "A",
      "direction": "동일한 리버스 와이드 축. 김지후(좌 중경)는 고개를 아래로 기울여 전경 중앙의 오열하는 강민숙을 응시하고, 정인우(우 중경)도 시선이 아래 중앙의 강민숙으로 수렴한다. 둘 다 렌즈를 보지 않음. 쇠사슬은 좌측 전경 바닥에서 뻗어 강민숙의 등을 타고 목덜미로 이어져 동굴 좌측 구석으로 향한다.",
      "built_space": "좌우 암벽과 중앙 깊은 어둠, 자갈·인골이 깔린 바닥, 우측 하단의 젖은 피웅덩이 — 이전 스틸의 장소·재질·손전등 조명 분위기 유지. 강민숙은 전경 하단 중앙에서 웅크려 등을 보이며, 등·허리·엉덩이·다리(흙 묻은 바지)까지 인체 비율이 자연스럽게 연결됨.",
      "entities": "김지후: 캡, 갈색 재킷, 올리브 헨리넥, 카고팬츠, 부츠, 목걸이 없음 — 참조 일치. 정인우: 녹색 코트, 버건디 터틀넥, 브로치, 우측 팔 캐스트와 흰 슬링 — 참조 및 캐리드 스테이트 일치. 강민숙: 뒷모습, 헝클어진 회백색 머리, 피 묻은 낡은 니트, 목·등을 감은 녹슨 사슬, 하단에 더러운 바지 — 이전 스틸의 의상·사슬 상태와 일치.",
      "hard_violations": [],
      "physics": "두 구조자는 바닥에 두 발을 딛고 정지해 서 있으며 무게 이동이 자연스럽다. 강민숙은 무릎을 굽혀 바닥에 웅크린 자세로, 상체가 앞으로 숙여지고 하반신이 바닥에 닿아 지지된다 — 오열하며 몸을 접은 동작으로 설명 가능. 사슬은 바닥의 자갈 위로 늘어져 무게에 따라 처짐."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 8
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S71sh14"
  }
 },
 "S78sh8": {
  "input_fingerprint": "b7631c9cbbccb1bb",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 직접 추격에 나서겠다는 강영란(흰 머리 여자)의 굳건한 표정 클로즈업.\n\nLOCATION (lock): The boulder-side altar position within the nighttime ritual clearing. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: The dolly advances along 강영란(흰 머리 여자)'s eyeline and finishes at close facial distance, holding her in a firm three-quarter close-up rather than a flat frontal portrait. Her set jaw and unwavering eyes occupy most of the frame, with only a narrow, defocused trace of the blood-red rock remaining at the edge as contextual evidence.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 강영란(흰 머리 여자) in the middle-center of the frame, foreground, looks toward 고산들 offscreen.\n- KEY BACKGROUND ELEMENTS: blood-red rock (partially visible behind her); used as Reduced to a defocused edge shape that preserves continuity without competing with her expression.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Close, flickering bonfire light cuts hard planes across her face while retaining the scene's deep nighttime contrast.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same bonfire-lit clearing, bloodied rock, smoke, and red robe. Exclude the bowing man from the facial close-up; focus on the elder's firm decision to join the pursuit.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Youngran retains the red shaman robe as she resolves to join the pursuit.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강영란(흰 머리 여자) (Korean woman in her late 60s, neatly twisted white hair, graceful dignified face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "강영란은 프레임 왼쪽 중앙에 3/4 각도로 서서 시선을 화면 오른쪽으로 보내고 있고, 그 시선 끝에 오른쪽 전경으로 남자의 머리·어깨 뒷모습이 실제로 들어와 있다. 즉 오프스크린이어야 할 고산들이 화면 안에 들어온 오버더숄더 구성이 되어 버렸다.",
    "built_space": "야외 제단 공터. 왼쪽 가장자리에 붉은 바위의 흐릿한 조각, 오른쪽 뒤로 모닥불 화염이 디포커스로 보인다. 배경 구조물(제단 건물)은 안개·연기로 대체되어 참조 스틸의 목조 제단 정면이 확인되지 않는다.",
    "entities": "인물은 60대 후반 한국 여성, 백발을 뒤로 묶었고 붉은 무복에 흰 동정 — 캐릭터/이전 스틸과 일치. 다만 프롬프트가 '절하는 남자는 클로즈업에서 제외하라'고 명시했음에도 두 번째 인물(남자 뒷머리·어깨)이 프레임 안에 추가되어 있다.",
    "hard_violations": [
     "샷 텍스트·이전 스틸 사용 지침이 명시적으로 제외한 남성 인물이 프레임 오른쪽 전경에 실체로 등장(추가 인물)"
    ],
    "physics": "인물은 두 발로 선 상태의 흉부 위 클로즈업으로 지지에 문제 없음. 옷 주름, 불빛 반사, 피부 질감 모두 물리적으로 자연스럽다."
   },
   {
    "label": "B",
    "direction": "강영란은 화면 중앙에서 약간 오른쪽에 위치, 몸은 정면에 가깝고 얼굴은 살짝 좌측 3/4로 돌려 카메라 좌측 전방(오프스크린)을 응시. 시선 끝에 인물은 없어 고산들이 오프스크린이라는 조건은 지킨다.",
    "built_space": "왼쪽 가장자리에 붉은 바위, 그 옆 중경에 모닥불, 오른쪽 배경에 참조 스틸의 목조 제단 건물과 계단·촛불이 선명하게 보인다. 장소 연속성은 가장 잘 드러나나 배경이 상당히 또렷해 '흐릿한 가장자리 흔적만' 조건과 어긋난다.",
    "entities": "60대 후반 백발 한국 여성, 붉은 무복·흰 동정 — 정체성 일치. 추가 인물 없음. 다만 얼굴 생김새가 참조보다 다소 다르게(광대·눈매) 렌더링되었다.",
    "hard_violations": [],
    "physics": "선 자세의 상반신 프레임, 어깨·목 연결 자연스럽고 지지 문제 없음. 불꽃과 연기, 천의 주름 모두 실물감 있음."
   },
   {
    "label": "C",
    "direction": "강영란은 프레임 중앙(약간 우측)에 3/4 각도로 서서 시선을 화면 왼쪽 전방 오프스크린으로 단호하게 보낸다. 시선 끝에는 아무 인물도 없고 붉은 바위 쪽 공간이 비어 있어 '오프스크린 고산들을 본다'는 지시와 부합.",
    "built_space": "왼쪽 가장자리에 붉은 바위가 크게 디포커스된 형태로만 걸쳐 있고, 뒤쪽은 연기·모닥불 반사광의 깊은 어둠. 제단 공터의 밤·연기·불빛 무드가 이전 스틸과 이어지며 배경이 표정을 방해하지 않는다.",
    "entities": "60대 후반 한국 여성, 단정히 틀어올린 백발(뒷머리 쪽 쪽진 형태까지 보임), 붉은 무복에 흰 동정 — 캐릭터 레퍼런스 및 이전 스틸과 가장 근접. 추가 인물 없음.",
    "hard_violations": [],
    "physics": "선 자세의 얼굴 클로즈업, 목·어깨 연결과 옷 드레이프가 자연스럽고 공중에 떠 있는 요소 없음. 불빛이 얼굴 좌측면에 단단한 면을 만드는 광원 논리도 성립."
   }
  ],
  "totals": {
   "A": 3,
   "B": 6,
   "C": 9
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "'절하는 남자는 클로즈업에서 제외'라는 명시 지시를 어기고 남성 인물을 전경에 넣어 오버더숄더가 되어 버린 점이 결정적 하드 위반."
   },
   {
    "label": "B",
    "score": 6,
    "verdict_ko": "추가 인물 없이 오프스크린 시선은 지켰으나, 상반신까지 넓힌 미디엄 프레임에 제단 건물까지 또렷해 '얼굴이 화면 대부분을 차지하는 클로즈업 + 흐릿한 바위 가장자리만' 조건에서 벗어남."
   },
   {
    "label": "C",
    "score": 9,
    "verdict_ko": "얼굴이 프레임 대부분을 차지하는 3/4 클로즈업, 왼쪽 가장자리의 디포커스된 붉은 바위, 오프스크린을 향한 단호한 시선, 붉은 무복 유지까지 카메라·프레임 지시를 가장 정확히 충족."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S78sh4_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강영란(흰 머리 여자): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:772891>"
   }
  ],
  "critique": {
   "issues": []
  },
  "fix_skipped": true,
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S78sh4"
  }
 },
 "S81sh4::bgfirst_bg": {
  "input_fingerprint": "ee4ee052c7d3e14e",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 강민숙(홍은미)이 턱을 아래로 당긴 채 묵묵히 긍정의 뜻을 내비치는 순간의 클로즈업.\n\nLOCATION (lock): The concealed corner of the packed fishing-gear shed behind piled nets and equipment.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Complete the dolly-in at Kang Min-sook's eye level on a slightly off-frontal three-quarter angle, lowering the close framing just enough to retain the small descent of her chin. Her face occupies the left-center of frame, with only compressed traces of the crowded warehouse behind her.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 쌓인 어망 (창고 안에 층층이 쌓임); used as Soft compressed background that maintains the hidden warehouse context.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Dim, low-key warehouse illumination preserves restrained facial contrast and the weight of her silent confirmation.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 강민숙(홍은미)이 턱을 아래로 당긴 채 묵묵히 긍정의 뜻을 내비치는 순간의 클로즈업.\n\nLOCATION (lock): The concealed corner of the packed fishing-gear shed behind piled nets and equipment.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Complete the dolly-in at Kang Min-sook's eye level on a slightly off-frontal three-quarter angle, lowering the close framing just enough to retain the small descent of her chin. Her face occupies the left-center of frame, with only compressed traces of the crowded warehouse behind her.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 쌓인 어망 (창고 안에 층층이 쌓임); used as Soft compressed background that maintains the hidden warehouse context.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Dim, low-key warehouse illumination preserves restrained facial contrast and the weight of her silent confirmation.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S81sh4__bgfirst_bg.png",
  "asset_id": "6067a54f-1b1a-4403-b5aa-0c78f08b0b89",
  "input_asset_ids": [
   "e7ea7bb4-d837-4d0d-a0a0-7f7b5bb84d36",
   "1f5558ff-3316-495e-827f-db124264a781"
  ]
 },
 "S81sh4": {
  "input_fingerprint": "fc5321d4337bb726",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 강민숙(홍은미)이 턱을 아래로 당긴 채 묵묵히 긍정의 뜻을 내비치는 순간의 클로즈업.\n\nLOCATION (lock): The concealed corner of the packed fishing-gear shed behind piled nets and equipment. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Complete the dolly-in at Kang Min-sook's eye level on a slightly off-frontal three-quarter angle, lowering the close framing just enough to retain the small descent of her chin. Her face occupies the left-center of frame, with only compressed traces of the crowded warehouse behind her.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 쌓인 어망 (창고 안에 층층이 쌓임); used as Soft compressed background that maintains the hidden warehouse context.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Dim, low-key warehouse illumination preserves restrained facial contrast and the weight of her silent confirmation.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same cramped fishing-gear storehouse, stacked nets, traps, baskets, darkness, and blood-smeared fugitives. Exclude the hiding-behind-net composition from the close-up; show the mother quietly confirming the truth.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain coated in goat blood inside the gear shed. Jihu’s necklace is absent, and Inwoo’s injured arm remains casted.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강민숙(홍은미) (Korean woman in her late 40s, dark hair, refined narrow face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot inside a tight, built interior. The FIRST attached image (SHOT BACKGROUND) is the finished empty interior of this shot, and in a space this cramped its geometry is the truth of the shot — keep it EXACTLY: its camera, perspective, every panel, control, seat, mirror, window and fixture stay untouched, in the same place, at the same angle, in the same number.\n\nBefore you place anyone, count what the background shows: how many steering wheels or control surfaces, how many seats and which way each faces, where each mirror sits and what it could reflect from this camera. Those counts and placements are what you must still be able to make after the people are in. Adding a second rim, sliding a seat, turning a mirror or growing a new panel is a failure even when the person looks right.\n\nThe SECOND attached image (LAYOUT SKETCH) tells you where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Its background lines are not decoration — they are the same structure seen in line form, so use them to place each person correctly with respect to it: which seat the body occupies, which side of the wheel the hands are on, what the body passes in front of and what it passes behind. Where sketch and background disagree about the structure itself, the background wins.\n\nThe CHARACTER REFERENCE photographs show the real people.\n\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. A person may cover part of the structure — that is expected, and covering is not redrawing. What the body hides stays hidden; what remains visible stays exactly as the background had it. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 강민숙(홍은미)이 턱을 아래로 당긴 채 묵묵히 긍정의 뜻을 내비치는 순간의 클로즈업.\n\nLOCATION (lock): The concealed corner of the packed fishing-gear shed behind piled nets and equipment. The shot takes place here — the FIRST attached image (SHOT BACKGROUND) is this exact place, already built: its ground, structures, horizon, materials and lighting are the finished truth of this location and must not be redesigned or replaced. No location photograph is attached — read the place from that image alone, and add no scenery, structure, vehicle or fixture that it does not already show. This lock governs the place only; the figures in the shot follow the staging and pose instructions.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Complete the dolly-in at Kang Min-sook's eye level on a slightly off-frontal three-quarter angle, lowering the close framing just enough to retain the small descent of her chin. Her face occupies the left-center of frame, with only compressed traces of the crowded warehouse behind her.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 쌓인 어망 (창고 안에 층층이 쌓임); used as Soft compressed background that maintains the hidden warehouse context.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Dim, low-key warehouse illumination preserves restrained facial contrast and the weight of her silent confirmation.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain coated in goat blood inside the gear shed. Jihu’s necklace is absent, and Inwoo’s injured arm remains casted.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강민숙(홍은미) (Korean woman in her late 40s, dark hair, refined narrow face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 강민숙(홍은미)이 턱을 아래로 당긴 채 묵묵히 긍정의 뜻을 내비치는 순간의 클로즈업.\n\nLOCATION (lock): The concealed corner of the packed fishing-gear shed behind piled nets and equipment. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Complete the dolly-in at Kang Min-sook's eye level on a slightly off-frontal three-quarter angle, lowering the close framing just enough to retain the small descent of her chin. Her face occupies the left-center of frame, with only compressed traces of the crowded warehouse behind her.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 쌓인 어망 (창고 안에 층층이 쌓임); used as Soft compressed background that maintains the hidden warehouse context.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Dim, low-key warehouse illumination preserves restrained facial contrast and the weight of her silent confirmation.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same cramped fishing-gear storehouse, stacked nets, traps, baskets, darkness, and blood-smeared fugitives. Exclude the hiding-behind-net composition from the close-up; show the mother quietly confirming the truth.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain coated in goat blood inside the gear shed. Jihu’s necklace is absent, and Inwoo’s injured arm remains casted.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강민숙(홍은미) (Korean woman in her late 40s, dark hair, refined narrow face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S81sh4__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement AND the structure they sit inside — the sketched panels, seats, controls and openings are the same ones the background photograph shows, drawn as lines; read them to place each body correctly against that structure, and where the two disagree about the structure itself the background photograph wins)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S81sh4.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:862549>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L41B03.png"
    },
    {
     "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:862549>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "턱을 당긴 연기와 피 묻은 상태는 좋으나, 창고 전체가 넓게 드러나는 미디엄 프레임이라 '클로즈업 + 배경은 압축된 흔적만' 지시를 어겼고 노출된 전구가 저조도 무드를 깨뜨린다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "눈높이·약간 오프프론탈 3/4, 좌중앙 얼굴 배치, 배경을 압축된 흔적으로만 남긴 클로즈업과 턱을 내린 침묵의 긍정을 정확히 구현했고 인물 동일성·저조도 창고 조명도 일치한다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.417,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.417,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "턱을 당긴 연기와 피 묻은 상태는 좋으나, 창고 전체가 넓게 드러나는 미디엄 프레임이라 '클로즈업 + 배경은 압축된 흔적만' 지시를 어겼고 노출된 전구가 저조도 무드를 깨뜨린다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "눈높이·약간 오프프론탈 3/4, 좌중앙 얼굴 배치, 배경을 압축된 흔적으로만 남긴 클로즈업과 턱을 내린 침묵의 긍정을 정확히 구현했고 인물 동일성·저조도 창고 조명도 일치한다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "인물은 프레임 왼쪽-중앙에 위치하고 얼굴은 화면 오른쪽 아래를 향해 약간 3/4로 돌아가 있으며, 시선은 아래로 내려 앞쪽 바닥 방향에 떨어진다. 턱이 아래로 당겨진 미세한 끄덕임 상태가 눈꺼풀과 목선에서 읽힌다. 카메라 축은 대략 눈높이, 약간 오프-프론탈 3/4로 샷 지시와 일치.",
      "built_space": "배경은 창고 내부: 오른쪽에 층층이 포개진 플라스틱 바구니 더미와 금속 봉/막대, 왼쪽 뒤로 어망·통발 틀이 압축되어 보인다. 레퍼런스 로케이션의 적층 바구니와 통발 벽이 그대로 대응하며, 압축된 얕은 심도로 배경이 전경을 침범하지 않는다. 중복 구조물이나 불가능한 반사 없음.",
      "entities": "40대 후반 한국 여성, 검은 머리, 갸름하고 정제된 얼굴 — 캐릭터 레퍼런스와 골격·헤어 길이가 상당히 부합. 목과 턱선에 마른 어두운 붉은 특수분장 자국이 있어 '염소 피에 뒤덮인 상태' 유지. 어두운 셔츠 착용. 다른 인물 없음, 문자·워터마크 없음.",
      "hard_violations": [],
      "physics": "인물은 서 있는 상태로 어깨·목이 자연스럽게 중력에 놓여 있고, 머리카락이 아래로 드리워지며 옷이 어깨선에서 접힌다. 공중에 뜬 물체나 지지 없는 신체 없음. 배경 바구니 더미는 바닥에서 쌓여 서로 지지."
     },
     {
      "label": "B",
      "direction": "인물은 프레임 중앙~약간 왼쪽, 얼굴은 오른쪽으로 3/4 돌려 아래를 내려다본다. 턱 당김은 읽히나 시선 방향은 화면 오른쪽 아래. 카메라는 눈높이보다 살짝 높고, 무엇보다 프레임이 넓어 상반신+창고 전체가 들어와 클로즈업이 아니라 미디엄 샷에 가깝다.",
      "built_space": "창고 벽면 통발 격자, 적층 바구니, 금속 봉, 왼쪽 어망, 오른쪽 철망 바구니 등 레퍼런스 로케이션 요소가 거의 그대로 넓게 재현됨. 천장에 밝은 전구 하나가 노출되어 저조도 무드가 다소 약화. 구조물 중복이나 불가능한 반사는 없음.",
      "entities": "한국 여성이나 인상은 레퍼런스보다 젊고 얼굴형·이목구비 일치도가 낮은 편. 얼굴과 상의에 광범위한 붉은 특수분장 얼룩이 있어 '피에 뒤덮인 상태'는 충족. 밝은 니트 상의. 추가 인물 없음, 문자 없음.",
      "hard_violations": [],
      "physics": "서 있는 자세로 어깨·팔이 중력에 맞게 놓이고 옷 주름이 자연스러움. 떠 있는 물체 없음. 배경 집기들은 바닥·벽에 지지됨."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "눈높이·약간 오프-프론탈 3/4의 진짜 클로즈업으로 얼굴이 좌중앙에 놓이고 배경은 압축된 흔적만 남아 카메라 지시를 정확히 이행했으며, 턱을 당긴 침묵의 수긍과 저조도·특수분장 상태도 부합."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "연기와 로케이션 재현은 무난하나 상반신·창고 전경이 다 들어간 미디엄 프레이밍으로 '클로즈업' 지시를 어겼고, 노출된 전구가 저조도 무드를 약화시키며 얼굴 동일성도 레퍼런스와 덜 맞는다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "눈높이·약간 오프-프론탈 3/4의 진짜 클로즈업으로 얼굴이 좌중앙에 놓이고 배경은 압축된 흔적만 남아 카메라 지시를 정확히 이행했으며, 턱을 당긴 침묵의 수긍과 저조도·특수분장 상태도 부합."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "연기와 로케이션 재현은 무난하나 상반신·창고 전경이 다 들어간 미디엄 프레이밍으로 '클로즈업' 지시를 어겼고, 노출된 전구가 저조도 무드를 약화시키며 얼굴 동일성도 레퍼런스와 덜 맞는다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "인물은 프레임 왼쪽-중앙에 위치하고 얼굴은 화면 오른쪽 아래를 향해 약간 3/4로 돌아가 있으며, 시선은 아래로 내려 앞쪽 바닥 방향에 떨어진다. 턱이 아래로 당겨진 미세한 끄덕임 상태가 눈꺼풀과 목선에서 읽힌다. 카메라 축은 대략 눈높이, 약간 오프-프론탈 3/4로 샷 지시와 일치.",
      "built_space": "배경은 창고 내부: 오른쪽에 층층이 포개진 플라스틱 바구니 더미와 금속 봉/막대, 왼쪽 뒤로 어망·통발 틀이 압축되어 보인다. 레퍼런스 로케이션의 적층 바구니와 통발 벽이 그대로 대응하며, 압축된 얕은 심도로 배경이 전경을 침범하지 않는다. 중복 구조물이나 불가능한 반사 없음.",
      "entities": "40대 후반 한국 여성, 검은 머리, 갸름하고 정제된 얼굴 — 캐릭터 레퍼런스와 골격·헤어 길이가 상당히 부합. 목과 턱선에 마른 어두운 붉은 특수분장 자국이 있어 '염소 피에 뒤덮인 상태' 유지. 어두운 셔츠 착용. 다른 인물 없음, 문자·워터마크 없음.",
      "hard_violations": [],
      "physics": "인물은 서 있는 상태로 어깨·목이 자연스럽게 중력에 놓여 있고, 머리카락이 아래로 드리워지며 옷이 어깨선에서 접힌다. 공중에 뜬 물체나 지지 없는 신체 없음. 배경 바구니 더미는 바닥에서 쌓여 서로 지지."
     },
     {
      "label": "A",
      "direction": "인물은 프레임 중앙~약간 왼쪽, 얼굴은 오른쪽으로 3/4 돌려 아래를 내려다본다. 턱 당김은 읽히나 시선 방향은 화면 오른쪽 아래. 카메라는 눈높이보다 살짝 높고, 무엇보다 프레임이 넓어 상반신+창고 전체가 들어와 클로즈업이 아니라 미디엄 샷에 가깝다.",
      "built_space": "창고 벽면 통발 격자, 적층 바구니, 금속 봉, 왼쪽 어망, 오른쪽 철망 바구니 등 레퍼런스 로케이션 요소가 거의 그대로 넓게 재현됨. 천장에 밝은 전구 하나가 노출되어 저조도 무드가 다소 약화. 구조물 중복이나 불가능한 반사는 없음.",
      "entities": "한국 여성이나 인상은 레퍼런스보다 젊고 얼굴형·이목구비 일치도가 낮은 편. 얼굴과 상의에 광범위한 붉은 특수분장 얼룩이 있어 '피에 뒤덮인 상태'는 충족. 밝은 니트 상의. 추가 인물 없음, 문자 없음.",
      "hard_violations": [],
      "physics": "서 있는 자세로 어깨·팔이 중력에 맞게 놓이고 옷 주름이 자연스러움. 떠 있는 물체 없음. 배경 집기들은 바닥·벽에 지지됨."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 11,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "턱을 당긴 연기와 피 묻은 상태는 좋으나, 창고 전체가 넓게 드러나는 미디엄 프레임이라 '클로즈업 + 배경은 압축된 흔적만' 지시를 어겼고 노출된 전구가 저조도 무드를 깨뜨린다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "눈높이·약간 오프프론탈 3/4, 좌중앙 얼굴 배치, 배경을 압축된 흔적으로만 남긴 클로즈업과 턱을 내린 침묵의 긍정을 정확히 구현했고 인물 동일성·저조도 창고 조명도 일치한다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L41B03.png"
   },
   {
    "label": "CHARACTER REFERENCE — 강민숙(홍은미): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:862549>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "인물의 의상이 레퍼런스 이미지(회색 코트와 검은색 터틀넥)와 일치하지 않고 검은색 셔츠로 그려졌습니다.",
     "fix_en": "Replace the woman's black collared shirt with the dark grey wool coat worn open over a black turtleneck collar from the reference, preserving her face, her expression, the blood smears, her pose, the lighting, and the blurred warehouse background exactly as they are."
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the woman's black collared shirt with the dark grey wool coat worn open over a black turtleneck collar from the reference, preserving her face, her expression, the blood smears, her pose, the lighting, and the blurred warehouse background exactly as they are.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "프레이밍·시선·로케이션은 정확하고 피 얼룩도 더 뚜렷하지만, 의상이 캐릭터 레퍼런스의 회색 코트+터틀넥이 아닌 검은 셔츠로 바뀌어 있어 인물 워드로브 일치도가 떨어진다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 구도·턱 당긴 하향 시선·창고 락을 유지하면서 레퍼런스의 회색 울 코트와 검은 터틀넥을 되살렸고, 피 얼룩도 유지되어 워드로브 정합성에서 앞선다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.653,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.653,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "프레이밍·시선·로케이션은 정확하고 피 얼룩도 더 뚜렷하지만, 의상이 캐릭터 레퍼런스의 회색 코트+터틀넥이 아닌 검은 셔츠로 바뀌어 있어 인물 워드로브 일치도가 떨어진다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 구도·턱 당긴 하향 시선·창고 락을 유지하면서 레퍼런스의 회색 울 코트와 검은 터틀넥을 되살렸고, 피 얼룩도 유지되어 워드로브 정합성에서 앞선다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "턱 당긴 하향 시선의 클로즈업, 눈높이 3/4 각도, 좌중앙 배치, 압축된 창고 배경, 회색 코트+터틀넥의 의상 연속성과 목의 핏자국까지 계약을 모두 유지 — 변경이 가장 적다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "연기·프레이밍·로케이션은 A와 동등하나, 계약이 전혀 요구하지 않은 의상 변경(회색 코트+터틀넥 → 검은 오픈 셔츠)으로 캐릭터 레퍼런스 및 이전 스틸의 복장 연속성을 잃어 손실이 이득보다 크다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.639
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.639
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "턱 당긴 하향 시선의 클로즈업, 눈높이 3/4 각도, 좌중앙 배치, 압축된 창고 배경, 회색 코트+터틀넥의 의상 연속성과 목의 핏자국까지 계약을 모두 유지 — 변경이 가장 적다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "연기·프레이밍·로케이션은 A와 동등하나, 계약이 전혀 요구하지 않은 의상 변경(회색 코트+터틀넥 → 검은 오픈 셔츠)으로 캐릭터 레퍼런스 및 이전 스틸의 복장 연속성을 잃어 손실이 이득보다 크다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 13,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S81sh4__bgfirst_bg.png",
   "bg_asset_id": "6067a54f-1b1a-4403-b5aa-0c78f08b0b89",
   "bg_record_key": "S81sh4::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S80sh14"
  }
 },
 "S85sh3::bgfirst_bg": {
  "input_fingerprint": "17ad765c82e42953",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 고산들의 옆에서 한 손으로는 어구 창고를 가리키고, 다른 한 손은 허공에 펼쳐 들어 강하게 만류하는 자세로 굳어 있는 마을 청년 #1의 모습.\n\nLOCATION (lock): The village lane between the restaurant frontage and the pier-side fishing-gear shed.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track at eye level beside 고산들 without crossing the shared axis, placing 마을 청년 #1 in the near middle-right and 고산들 at the left edge while the 어구 창고 aligns beyond the youth's pointing arm. Hold the youth in a tense three-quarter profile, one arm indicating the structure and the opposite palm spread toward 고산들 as a forceful visual barrier.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 마을 청년 #1 in the middle-right of the frame, midground, points to 어구 창고; 어구 창고 in the upper-left of the frame, background; 고산들 in the middle-left of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선착장 옆 어구 창고 (선착장 옆에 위치함) — A camera-facing outer side is visible beyond the youth's pointing arm; used as The structure is aligned with the youth's extended hand, making the disputed search location immediately legible; 숲 쪽으로 이어지는 길 (고산들이 다시 올라가려던 방향); used as The path anchors 고산들's interrupted movement toward the forest.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight appropriate to the exterior setting keeps the figures and their directional gestures plainly readable.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 고산들의 옆에서 한 손으로는 어구 창고를 가리키고, 다른 한 손은 허공에 펼쳐 들어 강하게 만류하는 자세로 굳어 있는 마을 청년 #1의 모습.\n\nLOCATION (lock): The village lane between the restaurant frontage and the pier-side fishing-gear shed.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track at eye level beside 고산들 without crossing the shared axis, placing 마을 청년 #1 in the near middle-right and 고산들 at the left edge while the 어구 창고 aligns beyond the youth's pointing arm. Hold the youth in a tense three-quarter profile, one arm indicating the structure and the opposite palm spread toward 고산들 as a forceful visual barrier.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 마을 청년 #1 in the middle-right of the frame, midground, points to 어구 창고; 어구 창고 in the upper-left of the frame, background; 고산들 in the middle-left of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선착장 옆 어구 창고 (선착장 옆에 위치함) — A camera-facing outer side is visible beyond the youth's pointing arm; used as The structure is aligned with the youth's extended hand, making the disputed search location immediately legible; 숲 쪽으로 이어지는 길 (고산들이 다시 올라가려던 방향); used as The path anchors 고산들's interrupted movement toward the forest.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight appropriate to the exterior setting keeps the figures and their directional gestures plainly readable.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.\n\nThe THIRD attached image (STRUCTURE LOOK) is the identity source of the fixed structure at this location: its faces, openings, levels, materials and signage are truth. Where it conflicts with the LOCATION PHOTOGRAPH about the structure itself, the STRUCTURE LOOK wins; the photograph still governs the surroundings, time of day and lighting.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S85sh3__bgfirst_bg.png",
  "asset_id": "e50b2879-b96b-4a9b-8fd1-73e5f0ef8cef",
  "input_asset_ids": [
   "28727209-cb49-4d87-8d3b-6926a3d6bf68",
   "beccd038-3a97-4525-8c4a-25bfc35eda1c",
   "a724fb27-abad-4d06-bbe9-79bbb37368ab"
  ]
 },
 "S85sh3": {
  "input_fingerprint": "0b02718d38bb0150",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 고산들의 옆에서 한 손으로는 어구 창고를 가리키고, 다른 한 손은 허공에 펼쳐 들어 강하게 만류하는 자세로 굳어 있는 마을 청년 #1의 모습.\n\nLOCATION (lock): The village lane between the restaurant frontage and the pier-side fishing-gear shed. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track at eye level beside 고산들 without crossing the shared axis, placing 마을 청년 #1 in the near middle-right and 고산들 at the left edge while the 어구 창고 aligns beyond the youth's pointing arm. Hold the youth in a tense three-quarter profile, one arm indicating the structure and the opposite palm spread toward 고산들 as a forceful visual barrier.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 마을 청년 #1 in the middle-right of the frame, midground, points to 어구 창고; 어구 창고 in the upper-left of the frame, background; 고산들 in the middle-left of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선착장 옆 어구 창고 (선착장 옆에 위치함) — A camera-facing outer side is visible beyond the youth's pointing arm; used as The structure is aligned with the youth's extended hand, making the disputed search location immediately legible; 숲 쪽으로 이어지는 길 (고산들이 다시 올라가려던 방향); used as The path anchors 고산들's interrupted movement toward the forest.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight appropriate to the exterior setting keeps the figures and their directional gestures plainly readable.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain hidden inside the gear shed under their goat-blood disguise. Inwoo retains the .38 handgun and cast, while Jihu’s necklace remains on the ritual altar.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (Korean man in his mid-40s, closely cropped dark hair, cold angular face); 마을 청년 #1 (Korean young man, short dark hair, broad youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 고산들의 옆에서 한 손으로는 어구 창고를 가리키고, 다른 한 손은 허공에 펼쳐 들어 강하게 만류하는 자세로 굳어 있는 마을 청년 #1의 모습.\n\nLOCATION (lock): The village lane between the restaurant frontage and the pier-side fishing-gear shed. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track at eye level beside 고산들 without crossing the shared axis, placing 마을 청년 #1 in the near middle-right and 고산들 at the left edge while the 어구 창고 aligns beyond the youth's pointing arm. Hold the youth in a tense three-quarter profile, one arm indicating the structure and the opposite palm spread toward 고산들 as a forceful visual barrier.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 마을 청년 #1 in the middle-right of the frame, midground, points to 어구 창고; 어구 창고 in the upper-left of the frame, background; 고산들 in the middle-left of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선착장 옆 어구 창고 (선착장 옆에 위치함) — A camera-facing outer side is visible beyond the youth's pointing arm; used as The structure is aligned with the youth's extended hand, making the disputed search location immediately legible; 숲 쪽으로 이어지는 길 (고산들이 다시 올라가려던 방향); used as The path anchors 고산들's interrupted movement toward the forest.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight appropriate to the exterior setting keeps the figures and their directional gestures plainly readable.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain hidden inside the gear shed under their goat-blood disguise. Inwoo retains the .38 handgun and cast, while Jihu’s necklace remains on the ritual altar.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (Korean man in his mid-40s, closely cropped dark hair, cold angular face); 마을 청년 #1 (Korean young man, short dark hair, broad youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 고산들의 옆에서 한 손으로는 어구 창고를 가리키고, 다른 한 손은 허공에 펼쳐 들어 강하게 만류하는 자세로 굳어 있는 마을 청년 #1의 모습.\n\nLOCATION (lock): The village lane between the restaurant frontage and the pier-side fishing-gear shed. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nSTRUCTURE LOOK AUTHORITY: the attached STRUCTURE LOOK photograph is the identity of the fixed structure at this location — wherever that structure appears in the frame, its shape, proportions, openings, materials and colors are LOCKED to it. The LOCATION PHOTOGRAPH remains the authority for this shot's sub-space, surroundings, time of day and lighting. If the two conflict on the structure itself, the STRUCTURE LOOK photo wins; for everything else, the LOCATION PHOTOGRAPH wins.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Track at eye level beside 고산들 without crossing the shared axis, placing 마을 청년 #1 in the near middle-right and 고산들 at the left edge while the 어구 창고 aligns beyond the youth's pointing arm. Hold the youth in a tense three-quarter profile, one arm indicating the structure and the opposite palm spread toward 고산들 as a forceful visual barrier.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 마을 청년 #1 in the middle-right of the frame, midground, points to 어구 창고; 어구 창고 in the upper-left of the frame, background; 고산들 in the middle-left of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선착장 옆 어구 창고 (선착장 옆에 위치함) — A camera-facing outer side is visible beyond the youth's pointing arm; used as The structure is aligned with the youth's extended hand, making the disputed search location immediately legible; 숲 쪽으로 이어지는 길 (고산들이 다시 올라가려던 방향); used as The path anchors 고산들's interrupted movement toward the forest.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight appropriate to the exterior setting keeps the figures and their directional gestures plainly readable.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Minsuk, Jihu, and Inwoo remain hidden inside the gear shed under their goat-blood disguise. Inwoo retains the .38 handgun and cast, while Jihu’s necklace remains on the ritual altar.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (Korean man in his mid-40s, closely cropped dark hair, cold angular face); 마을 청년 #1 (Korean young man, short dark hair, broad youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S85sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S85sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:903959>"
    },
    {
     "label": "CHARACTER REFERENCE — 마을 청년 #1: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:976931>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L40B01.png"
    },
    {
     "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_wooden_restaurant_compound_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:903959>"
    },
    {
     "label": "CHARACTER REFERENCE — 마을 청년 #1: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:976931>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "청년은 화면 중앙우측에 서서 왼팔(카메라 기준 왼쪽)을 뻗어 화면 좌측 방향을 가리키는데, 손가락 끝이 향하는 곳은 어구 창고 본체가 아니라 그 오른쪽의 전신주/해안 도로 부근이다. 반대 손바닥은 카메라 쪽(우측 앞)으로 펼쳐져 있어 고산들 방향(좌측)이 아니라 오히려 카메라 렌즈 쪽을 막고 있다. 고산들은 좌측 가장자리에서 등을 보인 채 청년 쪽을 보고 있어 시선은 맞다. 즉 '창고를 가리키는 손'과 '고산들을 막는 손'의 방향이 서로 뒤바뀐 듯 어긋난다.",
      "built_space": "선착장 옆 어구 창고가 화면 좌측 중경에 있으나, 문이 닫힌 붉은 목재 문의 시멘트블록 건물로 렌더되어 LOCATION 사진의 열린 문 안 어구 선반 구성과 다르다. 우측의 흰 벽·기와지붕 가옥, 태양광 패널, 격자 미닫이 담, 붉은 대야는 위치 사진과 일치. 다만 카메라가 실내 문틀 밖으로 나와 있어 위치 사진의 서브스페이스는 대체로 유지된다.",
      "entities": "고산들: 40대 한국 남성, 비니, 올리브 조끼+체크 셔츠 — 레퍼런스와 일치하나 뒤통수/측면만 보인다. 마을 청년 #1: 젊은 한국 남성, 짧은 머리, 얼굴은 레퍼런스와 유사. 다만 의상이 1670년 조선 걸인 넝마 한복 그대로여서 현대 한국 어촌 장면의 마을 청년으로 읽히지 않는다(레퍼런스 의상 직역).",
      "physics": "두 인물 모두 두 발이 지면에 붙어 체중이 실려 있고, 뻗은 팔은 어깨에서 자연스럽게 연결된다. 공중에 뜬 물체 없음. 청년의 펼친 손바닥은 해부학적으로 정상.",
      "hard_violations": []
     },
     {
      "label": "B",
      "direction": "청년은 중앙우측에서 오른팔을 곧게 뻗어 화면 좌상단의 어구 창고(열린 문과 선반이 보이는 블록 건물)를 정확히 가리키고, 손가락 연장선이 창고 개구부에 떨어진다. 왼손은 손바닥을 세워 좌측의 고산들 쪽으로 펼쳐 제지하는 벽을 만든다. 시선도 고산들을 향한다. 고산들은 좌측에서 청년을 응시. 두 손의 지시 방향이 샷 텍스트와 정확히 일치.",
      "built_space": "어구 창고: 시멘트블록 벽, 물결 슬레이트 박공 지붕, 열린 철문 안쪽에 그릇·어구 선반 — LOCATION 사진과 형태·개구부·재질이 일치. 우측 흰 벽 가옥·태양광 지붕·격자 담·붉은 대야·호스·장대·구명환까지 위치 사진 요소가 제자리에 있다. 카메라가 식당 문틀(좌우 프레임 기둥) 안쪽 눈높이에서 잡혀 위치 사진의 서브스페이스를 유지.",
      "entities": "고산들: 40대, 짧은 각진 머리, 차가운 얼굴, 올리브 조끼+체크 셔츠 — 캐릭터 레퍼런스와 일치(모자는 없음). 마을 청년 #1: 짧은 검은 머리, 넓은 젊은 얼굴 — 얼굴은 레퍼런스와 대체로 유사하며 현대 어촌에 맞는 재킷·면바지 차림. 추가 인물 없음, 텍스트/워터마크 없음.",
      "physics": "두 인물 모두 발이 지면에 견고히 놓여 있고, 청년은 앞발에 체중을 실은 전진 자세로 팔을 뻗어 밀어내는 동작이 자연스럽게 성립. 공중에 떠 있는 신체나 물체 없음.",
      "hard_violations": []
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "가리키는 손끝이 어구 창고를 빗나가 전신주/해안 쪽을 향하고 제지하는 손바닥이 고산들이 아닌 카메라를 막고 있어 샷 텍스트의 두 방향이 모두 어긋나며, 창고 개구부도 닫힌 붉은 문으로 위치 사진과 불일치하고 청년 의상이 조선 넝마 그대로다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "한 손은 열린 어구 창고를 정확히 지시하고 다른 손바닥은 좌측 고산들을 향해 펼쳐 만류하는 지시 방향이 완벽하며, 위치 사진의 창고·가옥·격자 담·주간 광량까지 일치하는 아이레벨 와이드로 프레임 배치도 지정대로다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "가리키는 손끝이 어구 창고를 빗나가 전신주/해안 쪽을 향하고 제지하는 손바닥이 고산들이 아닌 카메라를 막고 있어 샷 텍스트의 두 방향이 모두 어긋나며, 창고 개구부도 닫힌 붉은 문으로 위치 사진과 불일치하고 청년 의상이 조선 넝마 그대로다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "한 손은 열린 어구 창고를 정확히 지시하고 다른 손바닥은 좌측 고산들을 향해 펼쳐 만류하는 지시 방향이 완벽하며, 위치 사진의 창고·가옥·격자 담·주간 광량까지 일치하는 아이레벨 와이드로 프레임 배치도 지정대로다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "청년은 화면 중앙우측에 서서 왼팔(카메라 기준 왼쪽)을 뻗어 화면 좌측 방향을 가리키는데, 손가락 끝이 향하는 곳은 어구 창고 본체가 아니라 그 오른쪽의 전신주/해안 도로 부근이다. 반대 손바닥은 카메라 쪽(우측 앞)으로 펼쳐져 있어 고산들 방향(좌측)이 아니라 오히려 카메라 렌즈 쪽을 막고 있다. 고산들은 좌측 가장자리에서 등을 보인 채 청년 쪽을 보고 있어 시선은 맞다. 즉 '창고를 가리키는 손'과 '고산들을 막는 손'의 방향이 서로 뒤바뀐 듯 어긋난다.",
      "built_space": "선착장 옆 어구 창고가 화면 좌측 중경에 있으나, 문이 닫힌 붉은 목재 문의 시멘트블록 건물로 렌더되어 LOCATION 사진의 열린 문 안 어구 선반 구성과 다르다. 우측의 흰 벽·기와지붕 가옥, 태양광 패널, 격자 미닫이 담, 붉은 대야는 위치 사진과 일치. 다만 카메라가 실내 문틀 밖으로 나와 있어 위치 사진의 서브스페이스는 대체로 유지된다.",
      "entities": "고산들: 40대 한국 남성, 비니, 올리브 조끼+체크 셔츠 — 레퍼런스와 일치하나 뒤통수/측면만 보인다. 마을 청년 #1: 젊은 한국 남성, 짧은 머리, 얼굴은 레퍼런스와 유사. 다만 의상이 1670년 조선 걸인 넝마 한복 그대로여서 현대 한국 어촌 장면의 마을 청년으로 읽히지 않는다(레퍼런스 의상 직역).",
      "physics": "두 인물 모두 두 발이 지면에 붙어 체중이 실려 있고, 뻗은 팔은 어깨에서 자연스럽게 연결된다. 공중에 뜬 물체 없음. 청년의 펼친 손바닥은 해부학적으로 정상.",
      "hard_violations": []
     },
     {
      "label": "B",
      "direction": "청년은 중앙우측에서 오른팔을 곧게 뻗어 화면 좌상단의 어구 창고(열린 문과 선반이 보이는 블록 건물)를 정확히 가리키고, 손가락 연장선이 창고 개구부에 떨어진다. 왼손은 손바닥을 세워 좌측의 고산들 쪽으로 펼쳐 제지하는 벽을 만든다. 시선도 고산들을 향한다. 고산들은 좌측에서 청년을 응시. 두 손의 지시 방향이 샷 텍스트와 정확히 일치.",
      "built_space": "어구 창고: 시멘트블록 벽, 물결 슬레이트 박공 지붕, 열린 철문 안쪽에 그릇·어구 선반 — LOCATION 사진과 형태·개구부·재질이 일치. 우측 흰 벽 가옥·태양광 지붕·격자 담·붉은 대야·호스·장대·구명환까지 위치 사진 요소가 제자리에 있다. 카메라가 식당 문틀(좌우 프레임 기둥) 안쪽 눈높이에서 잡혀 위치 사진의 서브스페이스를 유지.",
      "entities": "고산들: 40대, 짧은 각진 머리, 차가운 얼굴, 올리브 조끼+체크 셔츠 — 캐릭터 레퍼런스와 일치(모자는 없음). 마을 청년 #1: 짧은 검은 머리, 넓은 젊은 얼굴 — 얼굴은 레퍼런스와 대체로 유사하며 현대 어촌에 맞는 재킷·면바지 차림. 추가 인물 없음, 텍스트/워터마크 없음.",
      "physics": "두 인물 모두 발이 지면에 견고히 놓여 있고, 청년은 앞발에 체중을 실은 전진 자세로 팔을 뻗어 밀어내는 동작이 자연스럽게 성립. 공중에 떠 있는 신체나 물체 없음.",
      "hard_violations": []
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "청년은 화면 중앙우측에서 왼팔(카메라 기준 왼쪽)을 뻗어 검지로 화면 좌측 상단의 어구 창고(문이 열린 콘크리트 블록 창고, 선반과 어구가 보임)를 정확히 가리킨다. 시선도 같은 방향(고산들 쪽/창고 쪽)으로 향해 있다. 반대 손은 손바닥을 펴 고산들 쪽 허공으로 내밀어 제지 장벽을 만든다. 고산들은 화면 좌측 가장자리에서 청년을 향해 옆얼굴로 서 있고 시선은 청년/창고 방향. 두 지시 방향 모두 텍스트가 말한 대상에 착지한다.",
      "built_space": "장소는 레퍼런스와 동일한 어촌 골목: 좌측 콘크리트 블록 어구 창고(개방된 문, 선반의 통발), 그 뒤 소나무 숲과 숲으로 이어지는 길, 중앙 전신주와 스피커, 우측 흰 벽 슬레이트 지붕 가옥과 미닫이 목문, 하단 나무 격자문과 붉은 대야, 바다와 방파제, 구명환. 구조물 개수·배치 모두 1개씩으로 중복 없음. 전경 좌우의 창틀/문틀이 레퍼런스 시점의 실내 프레이밍을 일부 계승하지만 프레임을 나누는 패널은 아님.",
      "entities": "고산들: 40대 중반 한국 남성, 짧게 깎은 검은 머리, 각진 차가운 얼굴, 올리브 조끼+체크 셔츠 — 레퍼런스와 일치(비니 모자는 벗은 상태로 차이). 마을 청년 #1: 20대 초 한국 남성, 짧은 검은 머리, 넓고 앳된 얼굴 — 얼굴/체형 일치. 다만 의상이 현대 자켓+면 티+검정 팬츠로, 레퍼런스의 누더기 한복과 다름. 인물은 두 명뿐, 추가 인물 없음. 자막·워터마크 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 두 발이 지면에 닿아 체중이 실려 있고, 청년은 앞다리에 무게를 실은 전진 자세로 팔을 뻗어 제지 동작이 자연스럽다. 공중에 뜬 물체나 지지 없는 신체 없음. 그림자가 지면에 일관되게 떨어지고 낮의 자연광."
     },
     {
      "label": "B",
      "direction": "청년은 중앙우측에서 팔을 화면 좌측으로 뻗어 가리키는데, 손끝이 향하는 곳은 좌측 블록 건물(문이 닫힌 붉은 문 창고)이 아니라 그 오른쪽의 전신주/해안 바위·바다 쪽에 가깝다. 창고와 손끝의 정렬이 어긋난다. 다른 손은 손바닥을 펴 들었으나 고산들 쪽이 아니라 화면 우측(가옥 쪽, 고산들 반대 방향)으로 향해 있어 '고산들에게 내미는 제지 장벽'이 성립하지 않는다. 고산들은 좌측에서 뒷모습에 가까운 후측면으로 서 있어 얼굴이 거의 보이지 않는다.",
      "built_space": "동일 로케이션의 요소들(블록 창고, 전신주+스피커, 우측 흰 가옥, 나무 격자문, 붉은 대야, 구명환, 바다)이 모두 1개씩 배치되어 중복은 없다. 다만 어구 창고의 개구부가 레퍼런스(열린 철문+어구 선반)와 달리 붉은 목문이 닫힌 상태로 렌더되어 개구부 상태가 바뀌었고, 어구가 보이지 않아 '분쟁 장소'로서의 가독성이 약하다.",
      "entities": "고산들: 비니 모자, 올리브 조끼, 체크 셔츠 — 복장은 레퍼런스와 일치하나 뒤통수 위주라 얼굴 확인 어려움. 마을 청년 #1: 한국 젊은 남성, 짧은 검은 머리로 얼굴 일치하나 의상이 누더기 한복(1670 조선풍)으로 현대 어촌 배경과 충돌. 인물 두 명, 추가 인물·문자 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 지면에 두 발을 딛고 있으며 그림자도 정상. 공중 부양이나 지지 없는 물체 없음. 다만 청년의 두 팔이 좌우로 벌어진 대칭 포즈여서 만류 동작의 힘 전달이 약하다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "청년의 손끝이 좌상단 어구 창고에 정확히 정렬되고 반대 손바닥이 고산들 쪽으로 펼쳐져 만류 장벽을 이루며, 프레임 배치(청년 중앙우측·고산들 좌측·창고 좌상단 배경)와 낮 자연광·로케이션이 모두 일치한다; 청년 의상이 레퍼런스의 누더기 한복이 아닌 현대복인 점만 감점."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "가리키는 손이 어구 창고를 빗나가 전신주·바다 쪽을 향하고, 제지용 손바닥이 고산들 반대편(우측)으로 펼쳐져 '만류' 방향이 무너졌으며, 창고 개구부도 닫힌 붉은 문으로 바뀌어 지시 대상의 가독성이 약하다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "청년의 손끝이 좌상단 어구 창고에 정확히 정렬되고 반대 손바닥이 고산들 쪽으로 펼쳐져 만류 장벽을 이루며, 프레임 배치(청년 중앙우측·고산들 좌측·창고 좌상단 배경)와 낮 자연광·로케이션이 모두 일치한다; 청년 의상이 레퍼런스의 누더기 한복이 아닌 현대복인 점만 감점."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "가리키는 손이 어구 창고를 빗나가 전신주·바다 쪽을 향하고, 제지용 손바닥이 고산들 반대편(우측)으로 펼쳐져 '만류' 방향이 무너졌으며, 창고 개구부도 닫힌 붉은 문으로 바뀌어 지시 대상의 가독성이 약하다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "청년은 화면 중앙우측에서 왼팔(카메라 기준 왼쪽)을 뻗어 검지로 화면 좌측 상단의 어구 창고(문이 열린 콘크리트 블록 창고, 선반과 어구가 보임)를 정확히 가리킨다. 시선도 같은 방향(고산들 쪽/창고 쪽)으로 향해 있다. 반대 손은 손바닥을 펴 고산들 쪽 허공으로 내밀어 제지 장벽을 만든다. 고산들은 화면 좌측 가장자리에서 청년을 향해 옆얼굴로 서 있고 시선은 청년/창고 방향. 두 지시 방향 모두 텍스트가 말한 대상에 착지한다.",
      "built_space": "장소는 레퍼런스와 동일한 어촌 골목: 좌측 콘크리트 블록 어구 창고(개방된 문, 선반의 통발), 그 뒤 소나무 숲과 숲으로 이어지는 길, 중앙 전신주와 스피커, 우측 흰 벽 슬레이트 지붕 가옥과 미닫이 목문, 하단 나무 격자문과 붉은 대야, 바다와 방파제, 구명환. 구조물 개수·배치 모두 1개씩으로 중복 없음. 전경 좌우의 창틀/문틀이 레퍼런스 시점의 실내 프레이밍을 일부 계승하지만 프레임을 나누는 패널은 아님.",
      "entities": "고산들: 40대 중반 한국 남성, 짧게 깎은 검은 머리, 각진 차가운 얼굴, 올리브 조끼+체크 셔츠 — 레퍼런스와 일치(비니 모자는 벗은 상태로 차이). 마을 청년 #1: 20대 초 한국 남성, 짧은 검은 머리, 넓고 앳된 얼굴 — 얼굴/체형 일치. 다만 의상이 현대 자켓+면 티+검정 팬츠로, 레퍼런스의 누더기 한복과 다름. 인물은 두 명뿐, 추가 인물 없음. 자막·워터마크 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 두 발이 지면에 닿아 체중이 실려 있고, 청년은 앞다리에 무게를 실은 전진 자세로 팔을 뻗어 제지 동작이 자연스럽다. 공중에 뜬 물체나 지지 없는 신체 없음. 그림자가 지면에 일관되게 떨어지고 낮의 자연광."
     },
     {
      "label": "A",
      "direction": "청년은 중앙우측에서 팔을 화면 좌측으로 뻗어 가리키는데, 손끝이 향하는 곳은 좌측 블록 건물(문이 닫힌 붉은 문 창고)이 아니라 그 오른쪽의 전신주/해안 바위·바다 쪽에 가깝다. 창고와 손끝의 정렬이 어긋난다. 다른 손은 손바닥을 펴 들었으나 고산들 쪽이 아니라 화면 우측(가옥 쪽, 고산들 반대 방향)으로 향해 있어 '고산들에게 내미는 제지 장벽'이 성립하지 않는다. 고산들은 좌측에서 뒷모습에 가까운 후측면으로 서 있어 얼굴이 거의 보이지 않는다.",
      "built_space": "동일 로케이션의 요소들(블록 창고, 전신주+스피커, 우측 흰 가옥, 나무 격자문, 붉은 대야, 구명환, 바다)이 모두 1개씩 배치되어 중복은 없다. 다만 어구 창고의 개구부가 레퍼런스(열린 철문+어구 선반)와 달리 붉은 목문이 닫힌 상태로 렌더되어 개구부 상태가 바뀌었고, 어구가 보이지 않아 '분쟁 장소'로서의 가독성이 약하다.",
      "entities": "고산들: 비니 모자, 올리브 조끼, 체크 셔츠 — 복장은 레퍼런스와 일치하나 뒤통수 위주라 얼굴 확인 어려움. 마을 청년 #1: 한국 젊은 남성, 짧은 검은 머리로 얼굴 일치하나 의상이 누더기 한복(1670 조선풍)으로 현대 어촌 배경과 충돌. 인물 두 명, 추가 인물·문자 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 지면에 두 발을 딛고 있으며 그림자도 정상. 공중 부양이나 지지 없는 물체 없음. 다만 청년의 두 팔이 좌우로 벌어진 대칭 포즈여서 만류 동작의 힘 전달이 약하다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 8,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "readings": [
   {
    "label": "A",
    "direction": "청년은 화면 중앙우측에 서서 왼팔(카메라 기준 왼쪽)을 뻗어 화면 좌측 방향을 가리키는데, 손가락 끝이 향하는 곳은 어구 창고 본체가 아니라 그 오른쪽의 전신주/해안 도로 부근이다. 반대 손바닥은 카메라 쪽(우측 앞)으로 펼쳐져 있어 고산들 방향(좌측)이 아니라 오히려 카메라 렌즈 쪽을 막고 있다. 고산들은 좌측 가장자리에서 등을 보인 채 청년 쪽을 보고 있어 시선은 맞다. 즉 '창고를 가리키는 손'과 '고산들을 막는 손'의 방향이 서로 뒤바뀐 듯 어긋난다.",
    "built_space": "선착장 옆 어구 창고가 화면 좌측 중경에 있으나, 문이 닫힌 붉은 목재 문의 시멘트블록 건물로 렌더되어 LOCATION 사진의 열린 문 안 어구 선반 구성과 다르다. 우측의 흰 벽·기와지붕 가옥, 태양광 패널, 격자 미닫이 담, 붉은 대야는 위치 사진과 일치. 다만 카메라가 실내 문틀 밖으로 나와 있어 위치 사진의 서브스페이스는 대체로 유지된다.",
    "entities": "고산들: 40대 한국 남성, 비니, 올리브 조끼+체크 셔츠 — 레퍼런스와 일치하나 뒤통수/측면만 보인다. 마을 청년 #1: 젊은 한국 남성, 짧은 머리, 얼굴은 레퍼런스와 유사. 다만 의상이 1670년 조선 걸인 넝마 한복 그대로여서 현대 한국 어촌 장면의 마을 청년으로 읽히지 않는다(레퍼런스 의상 직역).",
    "physics": "두 인물 모두 두 발이 지면에 붙어 체중이 실려 있고, 뻗은 팔은 어깨에서 자연스럽게 연결된다. 공중에 뜬 물체 없음. 청년의 펼친 손바닥은 해부학적으로 정상.",
    "hard_violations": []
   },
   {
    "label": "B",
    "direction": "청년은 중앙우측에서 오른팔을 곧게 뻗어 화면 좌상단의 어구 창고(열린 문과 선반이 보이는 블록 건물)를 정확히 가리키고, 손가락 연장선이 창고 개구부에 떨어진다. 왼손은 손바닥을 세워 좌측의 고산들 쪽으로 펼쳐 제지하는 벽을 만든다. 시선도 고산들을 향한다. 고산들은 좌측에서 청년을 응시. 두 손의 지시 방향이 샷 텍스트와 정확히 일치.",
    "built_space": "어구 창고: 시멘트블록 벽, 물결 슬레이트 박공 지붕, 열린 철문 안쪽에 그릇·어구 선반 — LOCATION 사진과 형태·개구부·재질이 일치. 우측 흰 벽 가옥·태양광 지붕·격자 담·붉은 대야·호스·장대·구명환까지 위치 사진 요소가 제자리에 있다. 카메라가 식당 문틀(좌우 프레임 기둥) 안쪽 눈높이에서 잡혀 위치 사진의 서브스페이스를 유지.",
    "entities": "고산들: 40대, 짧은 각진 머리, 차가운 얼굴, 올리브 조끼+체크 셔츠 — 캐릭터 레퍼런스와 일치(모자는 없음). 마을 청년 #1: 짧은 검은 머리, 넓은 젊은 얼굴 — 얼굴은 레퍼런스와 대체로 유사하며 현대 어촌에 맞는 재킷·면바지 차림. 추가 인물 없음, 텍스트/워터마크 없음.",
    "physics": "두 인물 모두 발이 지면에 견고히 놓여 있고, 청년은 앞발에 체중을 실은 전진 자세로 팔을 뻗어 밀어내는 동작이 자연스럽게 성립. 공중에 떠 있는 신체나 물체 없음.",
    "hard_violations": []
   }
  ],
  "totals": {
   "A": 8,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 4,
    "verdict_ko": "가리키는 손끝이 어구 창고를 빗나가 전신주/해안 쪽을 향하고 제지하는 손바닥이 고산들이 아닌 카메라를 막고 있어 샷 텍스트의 두 방향이 모두 어긋나며, 창고 개구부도 닫힌 붉은 문으로 위치 사진과 불일치하고 청년 의상이 조선 넝마 그대로다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "한 손은 열린 어구 창고를 정확히 지시하고 다른 손바닥은 좌측 고산들을 향해 펼쳐 만류하는 지시 방향이 완벽하며, 위치 사진의 창고·가옥·격자 담·주간 광량까지 일치하는 아이레벨 와이드로 프레임 배치도 지정대로다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its spatial layout, surroundings, fixed features, time of day and lighting mood are spatial truth; stage the moment inside this place. If a STRUCTURE LOOK photograph is also attached, that photo wins for the fixed structure itself — this photograph wins for everything around it. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L40B01.png"
   },
   {
    "label": "STRUCTURE LOOK — the confirmed photograph of the fixed structure at this location: wherever the structure appears in the frame, its shape, proportions, materials, colors and openings are LOCKED to this photo. Never copy its camera framing, time of day or lighting — the shot text and the LOCATION PHOTOGRAPH are the authorities for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_wooden_restaurant_compound_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903959>"
   },
   {
    "label": "CHARACTER REFERENCE — 마을 청년 #1: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:976931>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "마을 청년 #1이 캐릭터 레퍼런스에 명시된 낡은 조선 시대 전통 의상 대신 현대식 재킷과 바지를 입고 있습니다.",
     "fix_en": "Change the pointing youth's clothing from modern casual wear into the ragged, torn, muddy traditional Joseon-era garments shown in his character reference, exposing bare skin where torn. Maintain his exact face, hairstyle, pose, the other character Go San-deul in his current outfit, the background shed, and the lighting.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "프롬프트의 우선순위 규칙에 따라 어구 창고는 STRUCTURE LOOK 레퍼런스의 목조 건물 형태를 따라야 하나, LOCATION 레퍼런스의 콘크리트 건물로 잘못 렌더링되었습니다.",
     "fix_en": "Replace the concrete block texture of the background shed with weathered vertical wooden planks and a dark corrugated roof to match the materials of the STRUCTURE LOOK reference. Strictly preserve the shed's current size and footprint, the characters, their poses, the foreground elements, and the exact camera framing.",
     "needs_regeneration": true,
     "unfixable": false
    },
    {
     "issue_ko": "마을 청년 #1이 어구 창고를 가리키는 오른손의 손가락 형태가 뭉개져 있고 비율이 왜곡되어 있습니다.",
     "fix_en": "",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 5,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the pointing youth's clothing from modern casual wear into the ragged, torn, muddy traditional Joseon-era garments shown in his character reference, exposing bare skin where torn. Maintain his exact face, hairstyle, pose, the other character Go San-deul in his current outfit, the background shed, and the lighting.\n- Replace the concrete block texture of the background shed with weathered vertical wooden planks and a dark corrugated roof to match the materials of the STRUCTURE LOOK reference. Strictly preserve the shed's current size and footprint, the characters, their poses, the foreground elements, and the exact camera framing.\n- \n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "손가락이 로케이션 사진과 동일한 콘크리트 어구 창고에 정확히 얹히고 반대 손바닥은 고산들을 막는 장벽으로 서며, 창고 재질·개구부가 로케이션 락을 그대로 유지 — 청년의 현대 복장이 캐릭터 레퍼런스 의상과 다른 점만 감점."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "청년 의상을 레퍼런스대로 되돌린 이득은 있으나, 요청하지 않은 어구 창고 재질·형태(콘크리트 블록→목조 판벽)를 바꿔 로케이션 고정 구조물 락을 훼손했고 현대 한국 어촌 씬에 조선 넝마가 놓여 시대 정합성까지 깨져 손실이 이득보다 큼."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.714,
      "B": 1.75
     },
     "adjusted": {
      "A": 1.714,
      "B": 1.75
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "B"
     },
     "agreed": false
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "손가락이 로케이션 사진과 동일한 콘크리트 어구 창고에 정확히 얹히고 반대 손바닥은 고산들을 막는 장벽으로 서며, 창고 재질·개구부가 로케이션 락을 그대로 유지 — 청년의 현대 복장이 캐릭터 레퍼런스 의상과 다른 점만 감점."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "청년 의상을 레퍼런스대로 되돌린 이득은 있으나, 요청하지 않은 어구 창고 재질·형태(콘크리트 블록→목조 판벽)를 바꿔 로케이션 고정 구조물 락을 훼손했고 현대 한국 어촌 씬에 조선 넝마가 놓여 시대 정합성까지 깨져 손실이 이득보다 큼."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "청년은 화면 우측 중간에 서서 왼팔(카메라 기준 왼쪽)을 뻗어 화면 좌측 상단의 어구 창고 개구부 쪽을 정확히 가리킨다. 다른 손은 손바닥을 펴 좌측의 고산들 쪽으로 세워 제지하는 벽을 만든다. 고산들은 좌측 가장자리에서 우측 청년 쪽으로 시선을 두고 있다. 가리키는 손끝 연장선이 창고 문 개구부에 닿아 지시 대상이 명확하다.",
      "built_space": "어구 창고는 좌측 상단 배경에 위치하고 카메라 정면 측면과 열린 문, 내부 선반과 그릇들이 보인다. 다만 창고 벽면이 로케이션 사진의 회색 시멘트 블록이 아니라 세로 널판 목재로 바뀌어 있어 로케이션·구조물 참조와 재질이 어긋난다. 우측 흰 벽 건물, 격자 미닫이문, 붉은 대야, 스피커 전신주, 방파제와 배 등 배경 요소는 로케이션과 일치. 인물 대비 배경 스케일은 정상.",
      "entities": "고산들: 40대 한국 남성, 짧은 머리, 각진 냉정한 얼굴, 올리브 조끼+체크 셔츠 — 참조와 일치(비니는 벗은 상태). 마을 청년 #1: 얼굴·헤어는 참조와 일치하는 젊은 한국 남성이나, 의상이 조선시대 걸레 같은 누더기 한복 차림이다. 현대 한국 어촌 마을 청년 장면인데 시대착오적 의상이며, 참조 인물 사진의 스튜디오 의상을 그대로 복사한 결과로 보인다. 추가 인물 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 두 발이 땅에 닿아 서 있고 무게중심이 안정적이다. 청년의 뻗은 팔은 어깨에서 자연스럽게 연결되며 옷자락은 중력에 맞게 늘어진다. 공중에 뜬 물체 없음."
     },
     {
      "label": "B",
      "direction": "청년은 우측 중간에서 상체를 앞으로 기울인 채 오른팔을 뻗어 화면 좌측 상단의 어구 창고 열린 문을 정확히 가리키고, 다른 손은 손바닥을 펴 고산들 쪽으로 세워 제지한다. 시선도 고산들을 향한다. 지시선이 창고 개구부에 정확히 착지.",
      "built_space": "어구 창고가 좌측 상단 배경에 있으며 로케이션 사진과 동일한 회색 시멘트 블록 벽체, 골슬레이트 지붕, 열린 철제 문, 내부 선반의 그릇들이 그대로 재현됨. 우측 흰 건물·격자문·붉은 대야·전신주 스피커·방파제·해안 배 등도 일치. 스케일 정상.",
      "entities": "고산들: 참조와 동일한 40대 남성, 올리브 조끼와 체크 셔츠. 마을 청년 #1: 참조의 얼굴·짧은 머리·넓은 얼굴형과 일치하며, 현대 한국 어촌에 어울리는 카키 재킷·베이지 티·검정 바지 차림 — 현대 배경 설정에 부합. 추가 인물 없음.",
      "hard_violations": [],
      "physics": "청년은 두 발을 앞뒤로 벌리고 무릎을 약간 굽혀 상체를 앞으로 내민 자세로, 앞발과 뒷발이 모두 지면에 닿아 팔을 뻗는 동작을 지지한다. 고산들도 안정적으로 서 있다. 부유 물체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "지시·제지 자세와 프레임 배치는 맞으나, 어구 창고 벽체를 로케이션의 시멘트 블록에서 목재 널판으로 바꿔 구조물 락을 훼손했고 청년에게 시대착오적 조선 누더기 의상을 입혀 '현대 한국 어촌' 설정과 충돌한다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "창고 지시선·제지 손바닥·좌우 배치·와이드 스케일이 모두 지시대로이고, 시멘트 블록 창고 등 로케이션 고정 요소를 그대로 지키며 청년의 현대 복장도 동시대 설정에 부합해 변경 손실이 가장 적다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "지시·제지 자세와 프레임 배치는 맞으나, 어구 창고 벽체를 로케이션의 시멘트 블록에서 목재 널판으로 바꿔 구조물 락을 훼손했고 청년에게 시대착오적 조선 누더기 의상을 입혀 '현대 한국 어촌' 설정과 충돌한다."
     },
     {
      "label": "A",
      "score": 9,
      "verdict_ko": "창고 지시선·제지 손바닥·좌우 배치·와이드 스케일이 모두 지시대로이고, 시멘트 블록 창고 등 로케이션 고정 요소를 그대로 지키며 청년의 현대 복장도 동시대 설정에 부합해 변경 손실이 가장 적다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "청년은 화면 우측 중간에 서서 왼팔(카메라 기준 왼쪽)을 뻗어 화면 좌측 상단의 어구 창고 개구부 쪽을 정확히 가리킨다. 다른 손은 손바닥을 펴 좌측의 고산들 쪽으로 세워 제지하는 벽을 만든다. 고산들은 좌측 가장자리에서 우측 청년 쪽으로 시선을 두고 있다. 가리키는 손끝 연장선이 창고 문 개구부에 닿아 지시 대상이 명확하다.",
      "built_space": "어구 창고는 좌측 상단 배경에 위치하고 카메라 정면 측면과 열린 문, 내부 선반과 그릇들이 보인다. 다만 창고 벽면이 로케이션 사진의 회색 시멘트 블록이 아니라 세로 널판 목재로 바뀌어 있어 로케이션·구조물 참조와 재질이 어긋난다. 우측 흰 벽 건물, 격자 미닫이문, 붉은 대야, 스피커 전신주, 방파제와 배 등 배경 요소는 로케이션과 일치. 인물 대비 배경 스케일은 정상.",
      "entities": "고산들: 40대 한국 남성, 짧은 머리, 각진 냉정한 얼굴, 올리브 조끼+체크 셔츠 — 참조와 일치(비니는 벗은 상태). 마을 청년 #1: 얼굴·헤어는 참조와 일치하는 젊은 한국 남성이나, 의상이 조선시대 걸레 같은 누더기 한복 차림이다. 현대 한국 어촌 마을 청년 장면인데 시대착오적 의상이며, 참조 인물 사진의 스튜디오 의상을 그대로 복사한 결과로 보인다. 추가 인물 없음.",
      "hard_violations": [],
      "physics": "두 인물 모두 두 발이 땅에 닿아 서 있고 무게중심이 안정적이다. 청년의 뻗은 팔은 어깨에서 자연스럽게 연결되며 옷자락은 중력에 맞게 늘어진다. 공중에 뜬 물체 없음."
     },
     {
      "label": "A",
      "direction": "청년은 우측 중간에서 상체를 앞으로 기울인 채 오른팔을 뻗어 화면 좌측 상단의 어구 창고 열린 문을 정확히 가리키고, 다른 손은 손바닥을 펴 고산들 쪽으로 세워 제지한다. 시선도 고산들을 향한다. 지시선이 창고 개구부에 정확히 착지.",
      "built_space": "어구 창고가 좌측 상단 배경에 있으며 로케이션 사진과 동일한 회색 시멘트 블록 벽체, 골슬레이트 지붕, 열린 철제 문, 내부 선반의 그릇들이 그대로 재현됨. 우측 흰 건물·격자문·붉은 대야·전신주 스피커·방파제·해안 배 등도 일치. 스케일 정상.",
      "entities": "고산들: 참조와 동일한 40대 남성, 올리브 조끼와 체크 셔츠. 마을 청년 #1: 참조의 얼굴·짧은 머리·넓은 얼굴형과 일치하며, 현대 한국 어촌에 어울리는 카키 재킷·베이지 티·검정 바지 차림 — 현대 배경 설정에 부합. 추가 인물 없음.",
      "hard_violations": [],
      "physics": "청년은 두 발을 앞뒤로 벌리고 무릎을 약간 굽혀 상체를 앞으로 내민 자세로, 앞발과 뒷발이 모두 지면에 닿아 팔을 뻗는 동작을 지지한다. 고산들도 안정적으로 서 있다. 부유 물체 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 17,
     "B": 12
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": false,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S85sh3__bgfirst_bg.png",
   "bg_asset_id": "e50b2879-b96b-4a9b-8fd1-73e5f0ef8cef",
   "bg_record_key": "S85sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate",
   "seed_attached": true
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "S86sh10": {
  "input_fingerprint": "51ab6bc7f0e30c40",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 고산들의 엄청난 주먹이 앞을 막아선 조 형사의 머리에 강하게 적중하여 충격이 전달되는 mid-impact 순간.\n\nLOCATION (lock): The open pier between the charging villagers and the police patrol boat. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Cut to head height beside 조 형사, perpendicular to 고산들's attack path, and freeze the exact instant the fist compresses into the side of 조 형사's head. Keep both faces and the striking arm within a tight lateral composition, with 고산들의 forward drive entering from one edge and 조 형사의 head beginning to snap toward the opposite edge.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 고산들 in the middle-left of the frame, foreground, moves toward 조 형사's head; 조 형사 in the middle-right of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: 몰려오는 마을 사람들 (선착장을 향해 달려오는 중); used as The advancing crowd remains softly legible behind the impact as continuing pressure toward the patrol boat; 선착장 (충돌이 벌어지는 장소); used as A narrow background strip fixes the collision at the pier rather than abstracting it into an isolated blow.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight renders the impact with hard visual clarity while the background falls away from the two profiles.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same pier, patrol boat, charging villagers, bright daylight, and ongoing escape. Exclude the boarding hands from the focal action; show the large man's fist striking the detective's head.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains casted aboard the patrol boat, and the .38 handgun stays in his possession. The fugitives retain the goat-blood coating.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (Korean man in his mid-40s, closely cropped dark hair, cold angular face); 조 형사 (Korean man in his 40s, short dark hair, approachable mature face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "고산들이 중앙 좌측에서 우측 조 형사를 향해 팔을 뻗고 있으나, 주먹이 머리(관자놀이·턱)가 아니라 조 형사의 목/턱 아래 인후부를 붙잡듯 밀고 있다. 즉 타격점이 '머리'가 아니다. 조 형사의 머리는 반대 가장자리로 꺾이기보다 뒤로 젖혀지며 위를 향해 있다. 두 사람 모두 프로필로 마주보는 구도는 맞고, 카메라는 공격 경로에 수직인 측면 눈높이.",
    "built_space": "선착장 콘크리트 바닥, 우측에 경비정 선체(흰 상부·감청 하부)와 구명환, 좌측에 창고 벽면이 보여 이전 스틸의 장소와 재질이 대체로 일치. 배의 크기·거리도 실제 스케일 범위 안.",
    "entities": "고산들: 40대 중반, 짧은 머리, 각진 냉정한 얼굴, 올리브 조끼+체크 셔츠 — 캐릭터 레퍼런스와 일치(비니는 없음). 조 형사: 40대 한국 남성, 짧은 머리이나 의상이 베이지 재킷/지퍼 점퍼로 레퍼런스의 삼베 저고리와 다름. 배경 마을 사람들은 피(염소피) 묻은 상태로 달려옴 — 조건 부합. 추가 인물·텍스트 없음.",
    "hard_violations": [],
    "physics": "고산들은 두 발이 지면에 있고 몸통이 전진하며 팔을 뻗은 상태로 지지 확실. 조 형사도 지면에 서 있고 상체가 뒤로 밀리는 반응이 있으나, 목을 잡힌 듯한 손 모양이라 '주먹의 압착 순간'으로 읽기 어렵다. 침 튀김 입자는 있음."
   },
   {
    "label": "B",
    "direction": "고산들이 좌측에서 팔을 뻗지만 그의 주먹은 조 형사의 머리 앞쪽이 아니라 목덜미/뒷목 뒤로 넘어가 있어 타격점이 후두부 뒤 공중에 걸쳐 있는 애매한 위치다. 조 형사는 카메라 쪽으로 몸을 돌린 3/4 각도라 '머리가 반대편으로 꺾이는' 방향성이 약하다. 두 사람 시선도 서로 마주치지 않고 둘 다 화면 오른쪽/앞을 본다.",
    "built_space": "선착장·부두 방파제·우측에 흰 난간과 배 일부가 보이며 장소는 유지. 배가 프레임 오른쪽 끝에 매우 작게만 걸려 있어 '좁은 배경 띠'는 충족.",
    "entities": "고산들: 레퍼런스 얼굴·조끼·체크 셔츠 일치. 조 형사: 삼베 저고리 계열 상의로 레퍼런스 의상에 가깝고 얼굴도 40대 한국 남성. 배경 마을 사람들은 피 자국이 거의 없어 염소피 코팅 지속 조건이 약함. 추가 인물 없음.",
    "hard_violations": [
     "고산들의 뻗은 팔이 조 형사의 목 뒤를 관통하듯 지나가 주먹이 머리 반대편에 걸쳐 보이는 해부학적/공간적으로 성립하기 어려운 배치"
    ],
    "physics": "두 사람 모두 지면에 서 있고 지지는 명확. 다만 고산들의 어깨-팔꿈치-주먹 라인이 조 형사의 목을 지나쳐 뒤로 뻗어 있어 타격 압착의 힘 전달이 성립하지 않고, 조 형사의 상체 반응(꺾임)도 보이지 않는다."
   },
   {
    "label": "C",
    "direction": "고산들이 중앙 좌측 전경에서 우측으로 곧게 주먹을 뻗어, 주먹이 조 형사의 얼굴 측면(광대·턱)에 정확히 파고들어 뺨살이 눌리고 얼굴이 오른쪽 가장자리로 밀려나는 mid-impact가 명확히 보인다. 고산들의 시선은 표적인 조 형사의 머리에 고정.",
    "built_space": "선착장 콘크리트, 좌측 창고 벽, 우측 상단에 경비정 흰 조타실과 난간이 작은 띠로 보임 — 이전 스틸의 장소·재질과 일치하고 배는 실제 스케일대로 배경에 머문다.",
    "entities": "고산들: 짧은 머리, 각진 냉정한 얼굴, 올리브 워크 조끼+체크 셔츠 — 레퍼런스 일치. 조 형사: 40대 한국 남성, 짧은 머리, 삼베/무명 저고리에 핏자국 — 조 형사 레퍼런스 의상과 가장 근접. 배경 마을 사람들은 피 묻은 옷으로 선착장을 향해 달려옴. 텍스트·추가 인물 없음.",
    "hard_violations": [],
    "physics": "고산들은 지면에 발을 딛고 어깨-팔꿈치-주먹이 일직선으로 전진, 조 형사도 지면에 서서 충격으로 머리와 상체가 밀려나는 반응. 살 눌림과 접촉점이 실제 타격을 설명하며 떠 있는 신체 없음."
   }
  ],
  "totals": {
   "C": 9,
   "A": 6,
   "B": 3
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "C",
    "score": 9,
    "verdict_ko": "주먹이 조 형사의 머리 측면에 실제로 압착되며 머리가 반대편으로 밀리는 mid-impact를 정확히 포착했고, 측면 눈높이 타이트 클로즈업·좌우 배치·배경의 달려오는 마을 사람들과 좁은 선착장/경비정 띠까지 모두 충족."
   },
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "구도·장소·군중은 좋으나 주먹이 머리가 아닌 목/인후부를 밀고 있어 '머리에 적중'이라는 핵심 동작이 어긋나고, 조 형사의 의상도 레퍼런스의 삼베 저고리와 다르다."
   },
   {
    "label": "B",
    "score": 3,
    "verdict_ko": "뻗은 팔이 조 형사의 목 뒤로 지나가 타격점이 머리 뒤 허공에 놓여 충격 전달이 성립하지 않고, 두 인물 모두 서로를 향하지 않아 방향성과 mid-impact가 모두 실패."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S86sh7_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903959>"
   },
   {
    "label": "CHARACTER REFERENCE — 조 형사: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:957998>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "고산들(왼쪽)이 캐릭터 레퍼런스에서 착용하고 있는 비니를 쓰지 않고 맨머리로 묘사되었습니다.",
     "fix_en": "Draw the tight dark-brown beanie from the reference image onto the left man's head, covering his hair. Preserve his face, his expression, his clothing, the right man, the background crowd, the boat, and the exact camera framing.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "조 형사(오른쪽)의 머리가 똑바로 서 있고 주먹이 볼에 표면적으로만 닿아 있어, 프롬프트가 지시한 '주먹이 머리에 강하게 적중하여 볼이 눌리고 고개가 반대편으로 꺾이기 시작하는 순간'이 전혀 표현되지 않았습니다.",
     "fix_en": "Redraw the right man's head tilting sharply to the right, away from the incoming punch, and redraw the left man's fist sinking deeply into the right man's cheek to show heavy impact compression. Preserve the camera angle, both men's bodies and clothing below the neck, the left man's face, the background crowd, the boat, and the existing lighting.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Draw the tight dark-brown beanie from the reference image onto the left man's head, covering his hair. Preserve his face, his expression, his clothing, the right man, the background crowd, the boat, and the exact camera framing.\n- Redraw the right man's head tilting sharply to the right, away from the incoming punch, and redraw the left man's fist sinking deeply into the right man's cheek to show heavy impact compression. Preserve the camera angle, both men's bodies and clothing below the neck, the left man's face, the background crowd, the boat, and the existing lighting.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레임·타격 방향·장소 락은 충족하나, 계약이 요구하지 않은 변경으로 고산들의 캐릭터 레퍼런스 비니를 벗기고 머리·얼굴 인상까지 바꿔 정체성/의상 항목에서 손실이 크다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 측면 구도와 주먹-머리 명중 순간을 유지하면서 고산들의 레퍼런스 비니와 복장을 정확히 지켰고, 배경 군중·선착장·순찰선 스케일과 주광도 이전 스틸과 합치해 변화가 가장 적으면서 계약 충실도가 높다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.625,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.625,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레임·타격 방향·장소 락은 충족하나, 계약이 요구하지 않은 변경으로 고산들의 캐릭터 레퍼런스 비니를 벗기고 머리·얼굴 인상까지 바꿔 정체성/의상 항목에서 손실이 크다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 측면 구도와 주먹-머리 명중 순간을 유지하면서 고산들의 레퍼런스 비니와 복장을 정확히 지켰고, 배경 군중·선착장·순찰선 스케일과 주광도 이전 스틸과 합치해 변화가 가장 적으면서 계약 충실도가 높다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "클로즈업 측면 구도, 주먹의 뺨 압박 mid-impact, 배경 군중과 선착장 스트립, 낮 자연광이 모두 충족되고 고산들의 캐릭터 레퍼런스 복장(비니 포함)까지 그대로 유지해 변경이 가장 적다."
     },
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "구도·타격·장소는 A와 사실상 동일하나, 계약이 요구하지 않은 비니 제거로 캐릭터 레퍼런스의 의상 상태를 바꾼 손실이 있어 A보다 뒤진다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.875,
      "B": 1.875
     },
     "adjusted": {
      "A": 1.875,
      "B": 1.875
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "B"
     },
     "agreed": false
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "클로즈업 측면 구도, 주먹의 뺨 압박 mid-impact, 배경 군중과 선착장 스트립, 낮 자연광이 모두 충족되고 고산들의 캐릭터 레퍼런스 복장(비니 포함)까지 그대로 유지해 변경이 가장 적다."
     },
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "구도·타격·장소는 A와 사실상 동일하나, 계약이 요구하지 않은 비니 제거로 캐릭터 레퍼런스의 의상 상태를 바꾼 손실이 있어 A보다 뒤진다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 13,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S86sh7"
  }
 },
 "S86sh13": {
  "input_fingerprint": "c04ee53e67f44a1b",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 순식간에 벌어진 끔찍한 광경에 충격을 받아 두 눈이 튀어나올 듯 커진 채 얼어붙은 신 형사의 얼굴 클로즈업.\n\nLOCATION (lock): The pier-side boarding area beside the patrol boat and the fallen detective. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Cut directly to eye height in front of 신 형사 and reduce the image to a face close-up, his widened eyes held near the upper third while his jaw and locked shoulders remain visible below. The frontal position functions as a psychological collision rather than an address to camera; his gaze is fixed just past the lens toward the off-screen killing.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 신 형사 in the middle-center of the frame, foreground.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight holds 신 형사's frozen features in stark contrast while the setting recedes beyond recognition.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same pier, patrol boat, pursuing mob, daylight, and the detective's nearby position. Exclude the impact itself from the close-up; focus on the younger detective's horrified reaction.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu, Minsuk, and Inwoo remain blood-coated aboard the patrol boat; Inwoo retains his cast and handgun, while Jihu’s necklace remains absent.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 신 형사 (Korean man in his 30s, short dark hair, alert angular face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 4,
   "C": 9
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "장소·의상 연속성과 클로즈업 스케일은 좋으나 시선이 렌즈를 정면 직시해 '렌즈를 살짝 지나 화면 밖'이라는 지시에서 벗어나고, 배경 군중이 카메라를 향해 정지 응시하듯 보여 배경이 물러나야 한다는 조명·심도 지시도 약함.  ★위반: [gpt] 쇼트 텍스트가 신 형사만 보이도록 제한하는데 배경에 여러 명의 추가 인물이 식별 가능하게 등장한다."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "이전 스틸에서 확립된 피 묻은 베이지 한복 대신 남색 재킷으로 의상 연속성이 깨졌고, 순찰정 등 장소 고정 특징이 거의 사라져 로케이션 락이 약하며 시선도 렌즈 직시에 가깝다.  ★위반: [gpt] 쇼트 텍스트가 신 형사만 보이도록 제한하는데 배경에 여러 명의 추가 인물이 식별 가능하게 등장한다."
   },
   {
    "label": "C",
    "score": 9,
    "verdict_ko": "정면 아이레벨 얼굴 클로즈업, 상단 1/3의 부릅뜬 눈, 렌즈를 살짝 비껴 화면 밖 참극을 응시하는 시선, 부두·순찰정·피 묻은 의상의 연속성까지 모두 충족한 최선의 컷."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S86sh10_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 신 형사: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:862922>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "[Critical] 이전 컷(PREVIOUS SHOT STILL)에 명확히 있던 신 형사의 오른쪽 뺨과 입가의 핏자국이 완전히 사라졌습니다. 이전 컷의 상태가 그대로 유지되어야 한다는 지시를 위반했습니다.",
     "fix_en": "Add the smeared blood marks back onto the character's right cheek and chin to match his state in the previous shot, strictly preserving his exact facial expression, widened eyes, neat hair, clothing, the natural daylight, and the out-of-focus boat and pier in the background."
    },
    {
     "issue_ko": "[Major] 한복 상의의 깃이 전통적인 방식(섶이 겹치는 형태)으로 묘사되지 않고, 가운데 봉제선이 있는 현대식 브이넥 셔츠처럼 잘못 그려졌습니다.",
     "fix_en": ""
    },
    {
     "issue_ko": "[Major] 우측 배경에 있는 보트의 선체 하단과 선실 벽면에 흐릿한 문자가 포함되어 있어 모든 텍스트를 배제하라는(No text) 지시를 위반했습니다.",
     "fix_en": ""
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the smeared blood marks back onto the character's right cheek and chin to match his state in the previous shot, strictly preserving his exact facial expression, widened eyes, neat hair, clothing, the natural daylight, and the out-of-focus boat and pier in the background.\n- \n- \n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "프레이밍·시선·장소 락은 충족하나, 이전 스틸에서 확인된 얼굴의 피와 상처가 완전히 사라져 '이전 스틸의 각 인물 상태 지속' 연속성 조항에서 손실이 발생함."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 구도·아이레벨·중앙 배치를 유지하면서 뺨의 찰과상과 입가 혈흔으로 직전 스틸의 부상·유혈 상태 연속성까지 지켜, 배경·의상·인물 동일성을 바꾸지 않은 채 요구 항목만 보강함."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.764,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.764,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "프레이밍·시선·장소 락은 충족하나, 이전 스틸에서 확인된 얼굴의 피와 상처가 완전히 사라져 '이전 스틸의 각 인물 상태 지속' 연속성 조항에서 손실이 발생함."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 구도·아이레벨·중앙 배치를 유지하면서 뺨의 찰과상과 입가 혈흔으로 직전 스틸의 부상·유혈 상태 연속성까지 지켜, 배경·의상·인물 동일성을 바꾸지 않은 채 요구 항목만 보강함."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "정면 아이레벨 얼굴 클로즈업, 눈은 상단 1/3, 턱과 굳은 어깨 노출, 시선은 렌즈를 비껴 화면 밖 참사로 향함 — 프레이밍 지시를 충족하고 이전 스틸의 부두·순찰정·피 묻은 얼굴과 의상 상태를 그대로 유지한다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "동일 구도에 눈을 더 크게 뜬 연기 강화는 얻었으나, 계약이 요구하지 않은 얼굴·의상 혈흔 제거로 '피투성이 상태 지속' 잠금을 훼손하고 피부 질감도 평탄해져 손실이 이득보다 크다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.639
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.639
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "정면 아이레벨 얼굴 클로즈업, 눈은 상단 1/3, 턱과 굳은 어깨 노출, 시선은 렌즈를 비껴 화면 밖 참사로 향함 — 프레이밍 지시를 충족하고 이전 스틸의 부두·순찰정·피 묻은 얼굴과 의상 상태를 그대로 유지한다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "동일 구도에 눈을 더 크게 뜬 연기 강화는 얻었으나, 계약이 요구하지 않은 얼굴·의상 혈흔 제거로 '피투성이 상태 지속' 잠금을 훼손하고 피부 질감도 평탄해져 손실이 이득보다 크다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 13,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S86sh10"
  }
 },
 "S87sh2::bgfirst_bg": {
  "input_fingerprint": "1c3754d49c174c3d",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 몸의 반이 바닷물에 빠진 상태에서 한 손으로 배의 가장자리를 짚고 거칠게 몸을 솟구쳐 오르는 mid-action 순간의 고산들의 기괴하고 야수적인 전신.\n\nLOCATION (lock): The outer rail and bow edge of the moving police patrol boat just off the island pier.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Crane upward from water level just outside the bow-side rail, briefly widening and tilting up so 고산들 (육혈귀 각성 상태)'s entire body remains visible as one hand anchors him and his lower body surges out of the water. Place the rail diagonally across the middle of the frame, with his torso cresting above it and his trailing legs below, preserving the continuous rise toward the deck.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 고산들 (육혈귀 각성 상태) in the middle-center of the frame, midground, moves toward patrol boat deck; bow-side rail in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선수 가장자리 레일 바 (고산들이 한 손으로 붙잡은 상태) — Its outer side faces camera and runs diagonally across the frame beneath 고산들's gripping hand; used as The rail is the physical pivot for the one-handed ascent and divides sea level from deck level; 바닷물 (고산들의 몸 절반이 빠졌다가 솟구쳐 오르는 중); used as The water around his lower body shows the violent rebound used to reach the deck; 순찰정 갑판 (고산들이 올라서려는 위치); used as The upper surface beyond the rail provides the landing destination.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime light catches the water-soaked figure against the sea while preserving the harsh contrast of his beastlike form.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 몸의 반이 바닷물에 빠진 상태에서 한 손으로 배의 가장자리를 짚고 거칠게 몸을 솟구쳐 오르는 mid-action 순간의 고산들의 기괴하고 야수적인 전신.\n\nLOCATION (lock): The outer rail and bow edge of the moving police patrol boat just off the island pier.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Crane upward from water level just outside the bow-side rail, briefly widening and tilting up so 고산들 (육혈귀 각성 상태)'s entire body remains visible as one hand anchors him and his lower body surges out of the water. Place the rail diagonally across the middle of the frame, with his torso cresting above it and his trailing legs below, preserving the continuous rise toward the deck.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 고산들 (육혈귀 각성 상태) in the middle-center of the frame, midground, moves toward patrol boat deck; bow-side rail in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선수 가장자리 레일 바 (고산들이 한 손으로 붙잡은 상태) — Its outer side faces camera and runs diagonally across the frame beneath 고산들's gripping hand; used as The rail is the physical pivot for the one-handed ascent and divides sea level from deck level; 바닷물 (고산들의 몸 절반이 빠졌다가 솟구쳐 오르는 중); used as The water around his lower body shows the violent rebound used to reach the deck; 순찰정 갑판 (고산들이 올라서려는 위치); used as The upper surface beyond the rail provides the landing destination.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime light catches the water-soaked figure against the sea while preserving the harsh contrast of his beastlike form.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S87sh2__bgfirst_bg.png",
  "asset_id": "d126da23-149d-4adc-b5c5-6a8420830ec1",
  "input_asset_ids": [
   "19d52f0f-6830-4fca-b4f5-22af0bdc0303",
   "35d85063-4381-4750-9660-eb8724f7d321"
  ]
 },
 "S87sh2": {
  "input_fingerprint": "6d0d6cfc7555217b",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 몸의 반이 바닷물에 빠진 상태에서 한 손으로 배의 가장자리를 짚고 거칠게 몸을 솟구쳐 오르는 mid-action 순간의 고산들의 기괴하고 야수적인 전신.\n\nLOCATION (lock): The outer rail and bow edge of the moving police patrol boat just off the island pier. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Crane upward from water level just outside the bow-side rail, briefly widening and tilting up so 고산들 (육혈귀 각성 상태)'s entire body remains visible as one hand anchors him and his lower body surges out of the water. Place the rail diagonally across the middle of the frame, with his torso cresting above it and his trailing legs below, preserving the continuous rise toward the deck.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 고산들 (육혈귀 각성 상태) in the middle-center of the frame, midground, moves toward patrol boat deck; bow-side rail in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선수 가장자리 레일 바 (고산들이 한 손으로 붙잡은 상태) — Its outer side faces camera and runs diagonally across the frame beneath 고산들's gripping hand; used as The rail is the physical pivot for the one-handed ascent and divides sea level from deck level; 바닷물 (고산들의 몸 절반이 빠졌다가 솟구쳐 오르는 중); used as The water around his lower body shows the violent rebound used to reach the deck; 순찰정 갑판 (고산들이 올라서려는 위치); used as The upper surface beyond the rail provides the landing destination.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime light catches the water-soaked figure against the sea while preserving the harsh contrast of his beastlike form.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu, Minsuk, and Inwoo remain goat-blood-coated; Inwoo still has his cast and .38 handgun. Sandel boards soaked with seawater.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (육혈귀 각성 상태) (Korean man in his mid-40s, closely cropped dark hair, cold angular face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 몸의 반이 바닷물에 빠진 상태에서 한 손으로 배의 가장자리를 짚고 거칠게 몸을 솟구쳐 오르는 mid-action 순간의 고산들의 기괴하고 야수적인 전신.\n\nLOCATION (lock): The outer rail and bow edge of the moving police patrol boat just off the island pier. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Crane upward from water level just outside the bow-side rail, briefly widening and tilting up so 고산들 (육혈귀 각성 상태)'s entire body remains visible as one hand anchors him and his lower body surges out of the water. Place the rail diagonally across the middle of the frame, with his torso cresting above it and his trailing legs below, preserving the continuous rise toward the deck.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 고산들 (육혈귀 각성 상태) in the middle-center of the frame, midground, moves toward patrol boat deck; bow-side rail in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선수 가장자리 레일 바 (고산들이 한 손으로 붙잡은 상태) — Its outer side faces camera and runs diagonally across the frame beneath 고산들's gripping hand; used as The rail is the physical pivot for the one-handed ascent and divides sea level from deck level; 바닷물 (고산들의 몸 절반이 빠졌다가 솟구쳐 오르는 중); used as The water around his lower body shows the violent rebound used to reach the deck; 순찰정 갑판 (고산들이 올라서려는 위치); used as The upper surface beyond the rail provides the landing destination.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime light catches the water-soaked figure against the sea while preserving the harsh contrast of his beastlike form.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu, Minsuk, and Inwoo remain goat-blood-coated; Inwoo still has his cast and .38 handgun. Sandel boards soaked with seawater.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (육혈귀 각성 상태) (Korean man in his mid-40s, closely cropped dark hair, cold angular face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 몸의 반이 바닷물에 빠진 상태에서 한 손으로 배의 가장자리를 짚고 거칠게 몸을 솟구쳐 오르는 mid-action 순간의 고산들의 기괴하고 야수적인 전신.\n\nLOCATION (lock): The outer rail and bow edge of the moving police patrol boat just off the island pier. The shot takes place here — the attached LOCATION STRUCTURE PHOTOGRAPH is the single authority for this exact place — its fixed structure and permanent site details are LOCKED to it. No separate location photograph exists for this place. Build everything else strictly from the location text above and the shot text; the layout sketch (when attached) governs framing and placement only, and the shot text governs time of day, lighting and action.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Crane upward from water level just outside the bow-side rail, briefly widening and tilting up so 고산들 (육혈귀 각성 상태)'s entire body remains visible as one hand anchors him and his lower body surges out of the water. Place the rail diagonally across the middle of the frame, with his torso cresting above it and his trailing legs below, preserving the continuous rise toward the deck.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 고산들 (육혈귀 각성 상태) in the middle-center of the frame, midground, moves toward patrol boat deck; bow-side rail in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선수 가장자리 레일 바 (고산들이 한 손으로 붙잡은 상태) — Its outer side faces camera and runs diagonally across the frame beneath 고산들's gripping hand; used as The rail is the physical pivot for the one-handed ascent and divides sea level from deck level; 바닷물 (고산들의 몸 절반이 빠졌다가 솟구쳐 오르는 중); used as The water around his lower body shows the violent rebound used to reach the deck; 순찰정 갑판 (고산들이 올라서려는 위치); used as The upper surface beyond the rail provides the landing destination.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime light catches the water-soaked figure against the sea while preserving the harsh contrast of his beastlike form.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Jihu, Minsuk, and Inwoo remain goat-blood-coated; Inwoo still has his cast and .38 handgun. Sandel boards soaked with seawater.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (육혈귀 각성 상태) (Korean man in his mid-40s, closely cropped dark hair, cold angular face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S87sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S87sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들 (육혈귀 각성 상태): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:990262>"
    }
   ],
   "B": [
    {
     "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_police_patrol_boat_deck_sel.png"
    },
    {
     "label": "CHARACTER REFERENCE — 고산들 (육혈귀 각성 상태): the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:990262>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "readings": [
     {
      "label": "A",
      "direction": "인물은 프레임 왼쪽/카메라 쪽을 향해 시선과 상체를 돌리고 있고, 오른손은 뒤쪽 레일 바를 잡고 있으나 몸의 진행 방향은 갑판(오른쪽 뒤)이 아니라 배에서 멀어지는 왼쪽 바깥 바다 쪽이다. 즉 '갑판을 향해 솟구쳐 오르는' 방향과 반대로 이동 중으로 읽힌다. 레일은 화면 중앙을 대각선으로 가로지르긴 하나 인물은 레일 바깥(선수 밖)에 위치.",
      "built_space": "순찰정 선수 우현측, 조타실 창문·경광등·POLICE 방패 엠블럼·검정 펜더·청색 선체 등 레퍼런스 요소가 부분적으로 재현. 다만 엠블럼 문자가 '정정 POLICE'로 왜곡되어 있고, 레일 바 구조가 선수 난간 형태와 어긋나며, 선수 갑판 상면이 거의 보이지 않아 '착지 목적지'로서의 갑판이 프레임에 약하게만 존재.",
      "entities": "고산들: 40대 중반 한국 남성으로 보이나 상의를 완전히 탈의한 근육질 상반신에 수영 반바지 차림 — 캐릭터 레퍼런스의 낚시복 복장이나 어떤 의복도 아님. 머리는 짧은 검은 머리로 부합. 바닷물 젖은 상태는 표현됨. 물보라와 스플래시는 실제 물성으로 보임.",
      "hard_violations": [
       "몸이 물 밖으로 거의 완전히 나와 있고 두 발이 수면 위 허공에 있으며, 지지하는 손(레일 그립)은 몸 뒤쪽에서 사실상 하중을 받지 못하는 각도 — 물 위를 달리는 듯한 무지지 부유 자세로 물리적으로 성립하지 않음"
      ],
      "physics": "'몸의 반이 바닷물에 빠진 상태'라는 지문과 달리 하반신 전체가 수면 위에 노출되어 있고, 발이 수면을 딛는 듯 스플래시만 있을 뿐 실제 지지면이 없다. 레일을 쥔 손은 몸 뒤편에 있어 상승 피벗이 되지 못하고, 발진 지점도 착지 지점도 없다. 결과적으로 수면 위를 질주하는 부유체로 읽힌다."
     },
     {
      "label": "B",
      "direction": "인물의 시선과 상체는 카메라/전방을 향하되 몸 전체는 선수 갑판(우측 상단) 쪽으로 밀어 올려지는 방향. 왼손이 선수 난간 바를 확실히 붙잡고 그 힘으로 몸을 갑판 위로 끌어올리는 방향성이 일치. 레일은 화면 중앙을 대각선으로 가로지르고 손은 그 위에 놓임.",
      "built_space": "레퍼런스 순찰정 구조(백색 조타실, 청색 갑판, 흑색 선체와 펜더, 백색 파이프 난간, '한강경찰대'·'KOREA POLICE'·경광등·상부 조명군·경찰 방패 엠블럼)가 거의 그대로 재현. 선수 갑판 상면이 착지 목적지로 명확히 보이고, 레일이 해수면과 갑판을 분할. 배경에 부두/섬이 있어 '섬 선착장 인근' 로케이션 텍스트와도 부합.",
      "entities": "고산들: 40대 중반 한국 남성, 짧게 깎은 검은 머리, 각진 차가운 얼굴 — 캐릭터 레퍼런스 인상과 부합. 다만 의복은 젖어 검게 변한 작업복/재킷으로 레퍼런스의 네이비·오렌지 낚시복 컬러는 아님(젖음으로 어둡게 읽힐 여지). 전신이 물에 젖은 상태 표현 양호.",
      "hard_violations": [],
      "physics": "왼손이 난간 바를 실제로 움켜쥐고 있고, 하반신은 선체 옆 물속/물보라 속에 잠겨 있으며 무릎이 선수 가장자리에 걸쳐 체중을 받는다. 손의 그립과 무릎 접촉이 명확한 지지점으로, 물에서 솟구쳐 갑판으로 올라오는 동작의 중간 프레임으로 성립. 스플래시가 몸 아래에서 튀어 반동 표현도 일치."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "레일을 쥔 손이 몸 뒤쪽에 있어 지지가 되지 않고 하반신이 완전히 수면 위 허공에 떠 물 위를 달리는 부유 자세 — '몸의 반이 물에 빠진 상태에서 한 손으로 레일을 짚고 솟구침'이라는 지문의 물리와 방향(갑판 쪽 상승)을 모두 놓쳤고, 상의 탈의 반바지 차림은 캐릭터 레퍼런스와도 어긋난다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "한 손이 선수 난간을 확실히 잡고 하반신은 물보라 속에 잠긴 채 갑판 쪽으로 솟구치는 순간을 지지점(그립·무릎)과 함께 정확히 구현했고, 레퍼런스 순찰정 구조·표식·주간 자연광·와이드 프레임과 대각선 레일 배치도 지시대로다."
     }
    ],
    "all_candidates_fail": false
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "레일을 쥔 손이 몸 뒤쪽에 있어 지지가 되지 않고 하반신이 완전히 수면 위 허공에 떠 물 위를 달리는 부유 자세 — '몸의 반이 물에 빠진 상태에서 한 손으로 레일을 짚고 솟구침'이라는 지문의 물리와 방향(갑판 쪽 상승)을 모두 놓쳤고, 상의 탈의 반바지 차림은 캐릭터 레퍼런스와도 어긋난다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "한 손이 선수 난간을 확실히 잡고 하반신은 물보라 속에 잠긴 채 갑판 쪽으로 솟구치는 순간을 지지점(그립·무릎)과 함께 정확히 구현했고, 레퍼런스 순찰정 구조·표식·주간 자연광·와이드 프레임과 대각선 레일 배치도 지시대로다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "인물은 프레임 왼쪽/카메라 쪽을 향해 시선과 상체를 돌리고 있고, 오른손은 뒤쪽 레일 바를 잡고 있으나 몸의 진행 방향은 갑판(오른쪽 뒤)이 아니라 배에서 멀어지는 왼쪽 바깥 바다 쪽이다. 즉 '갑판을 향해 솟구쳐 오르는' 방향과 반대로 이동 중으로 읽힌다. 레일은 화면 중앙을 대각선으로 가로지르긴 하나 인물은 레일 바깥(선수 밖)에 위치.",
      "built_space": "순찰정 선수 우현측, 조타실 창문·경광등·POLICE 방패 엠블럼·검정 펜더·청색 선체 등 레퍼런스 요소가 부분적으로 재현. 다만 엠블럼 문자가 '정정 POLICE'로 왜곡되어 있고, 레일 바 구조가 선수 난간 형태와 어긋나며, 선수 갑판 상면이 거의 보이지 않아 '착지 목적지'로서의 갑판이 프레임에 약하게만 존재.",
      "entities": "고산들: 40대 중반 한국 남성으로 보이나 상의를 완전히 탈의한 근육질 상반신에 수영 반바지 차림 — 캐릭터 레퍼런스의 낚시복 복장이나 어떤 의복도 아님. 머리는 짧은 검은 머리로 부합. 바닷물 젖은 상태는 표현됨. 물보라와 스플래시는 실제 물성으로 보임.",
      "hard_violations": [
       "몸이 물 밖으로 거의 완전히 나와 있고 두 발이 수면 위 허공에 있으며, 지지하는 손(레일 그립)은 몸 뒤쪽에서 사실상 하중을 받지 못하는 각도 — 물 위를 달리는 듯한 무지지 부유 자세로 물리적으로 성립하지 않음"
      ],
      "physics": "'몸의 반이 바닷물에 빠진 상태'라는 지문과 달리 하반신 전체가 수면 위에 노출되어 있고, 발이 수면을 딛는 듯 스플래시만 있을 뿐 실제 지지면이 없다. 레일을 쥔 손은 몸 뒤편에 있어 상승 피벗이 되지 못하고, 발진 지점도 착지 지점도 없다. 결과적으로 수면 위를 질주하는 부유체로 읽힌다."
     },
     {
      "label": "B",
      "direction": "인물의 시선과 상체는 카메라/전방을 향하되 몸 전체는 선수 갑판(우측 상단) 쪽으로 밀어 올려지는 방향. 왼손이 선수 난간 바를 확실히 붙잡고 그 힘으로 몸을 갑판 위로 끌어올리는 방향성이 일치. 레일은 화면 중앙을 대각선으로 가로지르고 손은 그 위에 놓임.",
      "built_space": "레퍼런스 순찰정 구조(백색 조타실, 청색 갑판, 흑색 선체와 펜더, 백색 파이프 난간, '한강경찰대'·'KOREA POLICE'·경광등·상부 조명군·경찰 방패 엠블럼)가 거의 그대로 재현. 선수 갑판 상면이 착지 목적지로 명확히 보이고, 레일이 해수면과 갑판을 분할. 배경에 부두/섬이 있어 '섬 선착장 인근' 로케이션 텍스트와도 부합.",
      "entities": "고산들: 40대 중반 한국 남성, 짧게 깎은 검은 머리, 각진 차가운 얼굴 — 캐릭터 레퍼런스 인상과 부합. 다만 의복은 젖어 검게 변한 작업복/재킷으로 레퍼런스의 네이비·오렌지 낚시복 컬러는 아님(젖음으로 어둡게 읽힐 여지). 전신이 물에 젖은 상태 표현 양호.",
      "hard_violations": [],
      "physics": "왼손이 난간 바를 실제로 움켜쥐고 있고, 하반신은 선체 옆 물속/물보라 속에 잠겨 있으며 무릎이 선수 가장자리에 걸쳐 체중을 받는다. 손의 그립과 무릎 접촉이 명확한 지지점으로, 물에서 솟구쳐 갑판으로 올라오는 동작의 중간 프레임으로 성립. 스플래시가 몸 아래에서 튀어 반동 표현도 일치."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "인물의 시선과 몸통이 프레임 왼쪽 아래(카메라 쪽)를 향하지만, 상체 상승 벡터와 무릎·발의 방향은 선수 갑판(오른쪽 위)으로 향한다. 한 손은 선수 레일 바를 실제로 움켜쥐고 있고, 그 손의 힘 방향이 몸을 갑판 쪽으로 끌어올리는 축과 일치한다. 레일은 프레임 중앙을 대각선으로 가로지르며 손 아래에 놓인다.",
      "built_space": "레퍼런스의 한강경찰대 순찰정 구조(백색 조타실, 청색 갑판, 흑색 선체, 경광등, 경찰 마크, '한강경찰대'·'KOREA POLICE' 표기, 선수 난간과 볼라드)가 거의 그대로 재현됨. 난간은 좌우 각 1세트로 중복 없음, 조타실은 하나, 창·문 배치도 레퍼런스와 일치. 배경에 섬과 부두(선착장)가 보여 '섬 선착장 인근' 로케이션 텍스트와 부합.",
      "entities": "40대 중반 한국 남성, 짧게 깎은 검은 머리, 각지고 차가운 얼굴 — 캐릭터 레퍼런스 인물과 얼굴·헤어·체격이 상당히 일치. 물에 흠뻑 젖은 어두운 상하의(전투복 계열)를 착용해 '바닷물에 절반이 잠겼다 솟구침' 상태와 부합. 추가 인물·발명된 오브젝트 없음. 텍스트/워터마크 없음.",
      "hard_violations": [],
      "physics": "왼손이 레일 바를 실제로 감아쥐어 상체 하중을 지지하고, 무릎/정강이가 선수 펜더 상단에 닿아 두 번째 지지점을 형성한다. 하반신 주변에 물이 격렬히 튀어 방금 물에서 솟구쳐 나온 리바운드가 보임. 공중 부양 요소 없음."
     },
     {
      "label": "B",
      "direction": "인물은 카메라 왼쪽(바다 바깥쪽)을 향해 달려 나가는 방향으로 몸이 향해 있어, 갑판으로 '올라가는' 상승 벡터가 아니라 배에서 멀어지는 수평 이동으로 읽힌다. 오른손이 레일을 잡고 있으나 그 힘의 방향은 몸을 갑판으로 끌어올리는 축과 반대다. 시선도 갑판이 아닌 전방 바다 쪽.",
      "built_space": "선체 측면(우현) 근접뷰로, 레퍼런스의 특징적 요소 상당수가 누락/변형: '한강경찰대'·'서울경찰청' 표기 없음, 경찰 마크의 글자가 '정 점'으로 왜곡된 변형 텍스트, 경광등이 청색 단일 돔으로 바뀜, 조타실 형태와 난간 배치가 레퍼런스와 다름. 섬 선착장 배경도 없음. 레일은 프레임 중앙 대각선으로 놓이긴 하나 선수 가장자리가 아닌 측면 난간으로 보임.",
      "entities": "상반신 탈의에 수영복 차림의 20~30대로 보이는 근육질 남성 — 캐릭터 레퍼런스의 40대 중반, 각지고 차가운 얼굴과 얼굴·나이·체형이 명확히 불일치. 의상 상태(젖은 낚시복/의복)도 부재. 경찰 마크 안 문자가 왜곡된 글자로 렌더링되어 텍스트 금지 조항과 로케이션 락 모두에 어긋남.",
      "hard_violations": [
       "레퍼런스 순찰정의 고정 표기가 왜곡된 문자('정 점')로 변형 렌더링 — 텍스트/마킹 왜곡"
      ],
      "physics": "오른손은 난간을 잡고 있고 왼발이 수면을 차고 있으나, 몸통이 거의 수평으로 배 바깥을 향해 뻗어 있어 '갑판으로 솟구쳐 오름'이 아니라 물 위를 달려 나가는 부자연스러운 자세. 상승 궤적의 착지점(갑판)이 없다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "레퍼런스 순찰정 구조·표기와 인물 동일성을 모두 지켰고, 한 손이 선수 레일을 짚고 하반신이 물보라와 함께 갑판 쪽으로 솟구치는 mid-action을 대각선 레일 구도로 정확히 구현했다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "인물이 캐릭터 레퍼런스와 다른 젊은 상반신 탈의 남성이고, 몸의 방향이 갑판이 아닌 바다 바깥으로 향해 '솟구쳐 오름'이 성립하지 않으며, 선체 표기가 왜곡 문자로 렌더링되어 로케이션 락도 깨졌다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "레퍼런스 순찰정 구조·표기와 인물 동일성을 모두 지켰고, 한 손이 선수 레일을 짚고 하반신이 물보라와 함께 갑판 쪽으로 솟구치는 mid-action을 대각선 레일 구도로 정확히 구현했다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "인물이 캐릭터 레퍼런스와 다른 젊은 상반신 탈의 남성이고, 몸의 방향이 갑판이 아닌 바다 바깥으로 향해 '솟구쳐 오름'이 성립하지 않으며, 선체 표기가 왜곡 문자로 렌더링되어 로케이션 락도 깨졌다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "인물의 시선과 몸통이 프레임 왼쪽 아래(카메라 쪽)를 향하지만, 상체 상승 벡터와 무릎·발의 방향은 선수 갑판(오른쪽 위)으로 향한다. 한 손은 선수 레일 바를 실제로 움켜쥐고 있고, 그 손의 힘 방향이 몸을 갑판 쪽으로 끌어올리는 축과 일치한다. 레일은 프레임 중앙을 대각선으로 가로지르며 손 아래에 놓인다.",
      "built_space": "레퍼런스의 한강경찰대 순찰정 구조(백색 조타실, 청색 갑판, 흑색 선체, 경광등, 경찰 마크, '한강경찰대'·'KOREA POLICE' 표기, 선수 난간과 볼라드)가 거의 그대로 재현됨. 난간은 좌우 각 1세트로 중복 없음, 조타실은 하나, 창·문 배치도 레퍼런스와 일치. 배경에 섬과 부두(선착장)가 보여 '섬 선착장 인근' 로케이션 텍스트와 부합.",
      "entities": "40대 중반 한국 남성, 짧게 깎은 검은 머리, 각지고 차가운 얼굴 — 캐릭터 레퍼런스 인물과 얼굴·헤어·체격이 상당히 일치. 물에 흠뻑 젖은 어두운 상하의(전투복 계열)를 착용해 '바닷물에 절반이 잠겼다 솟구침' 상태와 부합. 추가 인물·발명된 오브젝트 없음. 텍스트/워터마크 없음.",
      "hard_violations": [],
      "physics": "왼손이 레일 바를 실제로 감아쥐어 상체 하중을 지지하고, 무릎/정강이가 선수 펜더 상단에 닿아 두 번째 지지점을 형성한다. 하반신 주변에 물이 격렬히 튀어 방금 물에서 솟구쳐 나온 리바운드가 보임. 공중 부양 요소 없음."
     },
     {
      "label": "A",
      "direction": "인물은 카메라 왼쪽(바다 바깥쪽)을 향해 달려 나가는 방향으로 몸이 향해 있어, 갑판으로 '올라가는' 상승 벡터가 아니라 배에서 멀어지는 수평 이동으로 읽힌다. 오른손이 레일을 잡고 있으나 그 힘의 방향은 몸을 갑판으로 끌어올리는 축과 반대다. 시선도 갑판이 아닌 전방 바다 쪽.",
      "built_space": "선체 측면(우현) 근접뷰로, 레퍼런스의 특징적 요소 상당수가 누락/변형: '한강경찰대'·'서울경찰청' 표기 없음, 경찰 마크의 글자가 '정 점'으로 왜곡된 변형 텍스트, 경광등이 청색 단일 돔으로 바뀜, 조타실 형태와 난간 배치가 레퍼런스와 다름. 섬 선착장 배경도 없음. 레일은 프레임 중앙 대각선으로 놓이긴 하나 선수 가장자리가 아닌 측면 난간으로 보임.",
      "entities": "상반신 탈의에 수영복 차림의 20~30대로 보이는 근육질 남성 — 캐릭터 레퍼런스의 40대 중반, 각지고 차가운 얼굴과 얼굴·나이·체형이 명확히 불일치. 의상 상태(젖은 낚시복/의복)도 부재. 경찰 마크 안 문자가 왜곡된 글자로 렌더링되어 텍스트 금지 조항과 로케이션 락 모두에 어긋남.",
      "hard_violations": [
       "레퍼런스 순찰정의 고정 표기가 왜곡된 문자('정 점')로 변형 렌더링 — 텍스트/마킹 왜곡"
      ],
      "physics": "오른손은 난간을 잡고 있고 왼발이 수면을 차고 있으나, 몸통이 거의 수평으로 배 바깥을 향해 뻗어 있어 '갑판으로 솟구쳐 오름'이 아니라 물 위를 달려 나가는 부자연스러운 자세. 상승 궤적의 착지점(갑판)이 없다."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 6,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "readings": [
   {
    "label": "A",
    "direction": "인물은 프레임 왼쪽/카메라 쪽을 향해 시선과 상체를 돌리고 있고, 오른손은 뒤쪽 레일 바를 잡고 있으나 몸의 진행 방향은 갑판(오른쪽 뒤)이 아니라 배에서 멀어지는 왼쪽 바깥 바다 쪽이다. 즉 '갑판을 향해 솟구쳐 오르는' 방향과 반대로 이동 중으로 읽힌다. 레일은 화면 중앙을 대각선으로 가로지르긴 하나 인물은 레일 바깥(선수 밖)에 위치.",
    "built_space": "순찰정 선수 우현측, 조타실 창문·경광등·POLICE 방패 엠블럼·검정 펜더·청색 선체 등 레퍼런스 요소가 부분적으로 재현. 다만 엠블럼 문자가 '정정 POLICE'로 왜곡되어 있고, 레일 바 구조가 선수 난간 형태와 어긋나며, 선수 갑판 상면이 거의 보이지 않아 '착지 목적지'로서의 갑판이 프레임에 약하게만 존재.",
    "entities": "고산들: 40대 중반 한국 남성으로 보이나 상의를 완전히 탈의한 근육질 상반신에 수영 반바지 차림 — 캐릭터 레퍼런스의 낚시복 복장이나 어떤 의복도 아님. 머리는 짧은 검은 머리로 부합. 바닷물 젖은 상태는 표현됨. 물보라와 스플래시는 실제 물성으로 보임.",
    "hard_violations": [
     "몸이 물 밖으로 거의 완전히 나와 있고 두 발이 수면 위 허공에 있으며, 지지하는 손(레일 그립)은 몸 뒤쪽에서 사실상 하중을 받지 못하는 각도 — 물 위를 달리는 듯한 무지지 부유 자세로 물리적으로 성립하지 않음"
    ],
    "physics": "'몸의 반이 바닷물에 빠진 상태'라는 지문과 달리 하반신 전체가 수면 위에 노출되어 있고, 발이 수면을 딛는 듯 스플래시만 있을 뿐 실제 지지면이 없다. 레일을 쥔 손은 몸 뒤편에 있어 상승 피벗이 되지 못하고, 발진 지점도 착지 지점도 없다. 결과적으로 수면 위를 질주하는 부유체로 읽힌다."
   },
   {
    "label": "B",
    "direction": "인물의 시선과 상체는 카메라/전방을 향하되 몸 전체는 선수 갑판(우측 상단) 쪽으로 밀어 올려지는 방향. 왼손이 선수 난간 바를 확실히 붙잡고 그 힘으로 몸을 갑판 위로 끌어올리는 방향성이 일치. 레일은 화면 중앙을 대각선으로 가로지르고 손은 그 위에 놓임.",
    "built_space": "레퍼런스 순찰정 구조(백색 조타실, 청색 갑판, 흑색 선체와 펜더, 백색 파이프 난간, '한강경찰대'·'KOREA POLICE'·경광등·상부 조명군·경찰 방패 엠블럼)가 거의 그대로 재현. 선수 갑판 상면이 착지 목적지로 명확히 보이고, 레일이 해수면과 갑판을 분할. 배경에 부두/섬이 있어 '섬 선착장 인근' 로케이션 텍스트와도 부합.",
    "entities": "고산들: 40대 중반 한국 남성, 짧게 깎은 검은 머리, 각진 차가운 얼굴 — 캐릭터 레퍼런스 인상과 부합. 다만 의복은 젖어 검게 변한 작업복/재킷으로 레퍼런스의 네이비·오렌지 낚시복 컬러는 아님(젖음으로 어둡게 읽힐 여지). 전신이 물에 젖은 상태 표현 양호.",
    "hard_violations": [],
    "physics": "왼손이 난간 바를 실제로 움켜쥐고 있고, 하반신은 선체 옆 물속/물보라 속에 잠겨 있으며 무릎이 선수 가장자리에 걸쳐 체중을 받는다. 손의 그립과 무릎 접촉이 명확한 지지점으로, 물에서 솟구쳐 갑판으로 올라오는 동작의 중간 프레임으로 성립. 스플래시가 몸 아래에서 튀어 반동 표현도 일치."
   }
  ],
  "totals": {
   "A": 6,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 3,
    "verdict_ko": "레일을 쥔 손이 몸 뒤쪽에 있어 지지가 되지 않고 하반신이 완전히 수면 위 허공에 떠 물 위를 달리는 부유 자세 — '몸의 반이 물에 빠진 상태에서 한 손으로 레일을 짚고 솟구침'이라는 지문의 물리와 방향(갑판 쪽 상승)을 모두 놓쳤고, 상의 탈의 반바지 차림은 캐릭터 레퍼런스와도 어긋난다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "한 손이 선수 난간을 확실히 잡고 하반신은 물보라 속에 잠긴 채 갑판 쪽으로 솟구치는 순간을 지지점(그립·무릎)과 함께 정확히 구현했고, 레퍼런스 순찰정 구조·표식·주간 자연광·와이드 프레임과 대각선 레일 배치도 지시대로다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION STRUCTURE PHOTOGRAPH — the confirmed photograph of this exact place and its fixed structure: it is the SINGLE authority for the location, the structure's shape, proportions, materials, colors, openings and every permanent site detail. Never copy its camera framing, time of day or lighting — the shot text is the authority for those.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/background_chain/seed_bg_police_patrol_boat_deck_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들 (육혈귀 각성 상태): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:990262>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "카메라 앵글이 수면에서 올려다보는 구도가 아니며, 인물이 선체 밖 바닷물에서 솟구쳐 오르는 대신 이미 배 안쪽 구조물 위에 올라서 있습니다.",
     "fix_en": "Erase the deck and solid structure beneath the character's waist, replacing them with splashing seawater and his submerged trailing legs, to simulate him hanging over the outer edge. Preserve his face, upper torso, the boat cabin, and the background.",
     "needs_regeneration": true
    },
    {
     "issue_ko": "배 선수 가장자리 구조가 참조 사진의 개방형 파이프 레일이 아닌, 두꺼운 흰색 장갑벽 형태로 잘못 생성되었습니다.",
     "fix_en": "Change the solid white wall at the front of the boat into an open white tubular metal railing sitting on a flat blue deck. Preserve the character, the boat cabin behind him, and the sea background.",
     "needs_regeneration": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Erase the deck and solid structure beneath the character's waist, replacing them with splashing seawater and his submerged trailing legs, to simulate him hanging over the outer edge. Preserve his face, upper torso, the boat cabin, and the background.\n- Change the solid white wall at the front of the boat into an open white tubular metal railing sitting on a flat blue deck. Preserve the character, the boat cabin behind him, and the sea background.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "readings": [
     {
      "label": "A",
      "direction": "고산들의 시선과 상체는 프레임 좌하 전방(카메라 쪽)을 향하지만 몸통의 진행 방향은 레일을 넘어 갑판 쪽으로 솟구치는 방향으로 읽힌다. 왼손(화면 좌측 손)이 선수 레일 바를 실제로 움켜쥐고 있고, 반대 팔은 뒤로 뻗어 반동을 만든다. 뻗은 팔·머리·시선이 모두 갑판 상방으로 이어지는 상승선과 일치한다.",
      "built_space": "순찰정은 참조 사진의 구조와 일치한다 — 흰 조타실, 파란 갑판, 검은 선체, '한강경찰대'/'KOREA POLICE'/'한강경찰' 표기, 경찰 마크, 경광등, 조타실 상부 서치라이트 배열. 선수 좌현 난간이 별도로 서 있고, 인물이 잡은 레일은 선수 가장자리 바로 프레임 중앙을 대각으로 가로지른다. 레일 개수·배치가 참조와 큰 모순 없음. 다만 인물이 잡은 레일 하단부가 선체 상부 흰 판과 겹치며 접합이 다소 모호하다.",
      "entities": "인물은 40대 중반 한국 남성, 아주 짧게 깎은 검은 머리, 각진 차가운 얼굴 — 캐릭터 참조와 일치. 물에 흠뻑 젖은 어두운 군용/작업복 상하의로 야수적 인상을 유지. 추가 인물 없음. 텍스트·워터마크 없음. 낮의 자연광, 해면 반사 정상.",
      "hard_violations": [],
      "physics": "지지점이 명확하다: 한 손이 레일을 확실히 잡고, 하반신은 여전히 물 속/수면에 잠겨 있으며 발 주변에서 물보라가 튄다. 오른쪽 무릎·정강이가 선체 흰 면에 닿아 밀어올리는 접촉이 보인다. 물튀김이 다리 주변에 집중되어 솟구침의 반동과 일치. 공중에 떠 있는 부양체 없음."
     },
     {
      "label": "B",
      "direction": "시선·상체 방향은 A와 사실상 동일하게 좌하 전방. 손은 레일을 잡고 있으나 잡은 레일이 선수 좌현 난간에서 뽑혀 나온 듯 길게 휘어져 프레임 우측 아래로 이어지며, 그 레일 자체가 선체와 연결되지 않은 구간을 가진다 — 즉 그가 붙잡은 '레일'의 종착점이 불명확하다.",
      "built_space": "선박 구조는 참조와 대체로 일치하나 선수부가 A와 달리 재구성되었다: 좌현 난간이 선수 전체를 감싸는 곡면 파이프 난간으로 바뀌었고, 그 난간에서 다시 별개의 긴 파이프가 인물의 손 아래로 뻗어 우현 난간까지 이어지는 이중 레일 구조가 생겼다. 참조 사진의 선수 난간은 짧은 두 개의 독립 난간이며 선수를 감싸는 연속 곡선 레일은 없다 — 구조물이 증식·변형됨. 또한 선수 갑판 앞부분(검은 방현재/선수 곡선)이 잘려 나가 파란 갑판이 공중에 뜬 형태로 보인다.",
      "entities": "인물 동일 — 40대 중반, 짧은 머리, 각진 얼굴, 젖은 어두운 작업복. 참조와 일치. 추가 인물 없음. 선체 표기 '서울경찰청','한강경찰','한강경찰대' 가독. 텍스트 오염 없음.",
      "hard_violations": [
       "참조 구조에 없는 선수 감싸기 곡면 난간 + 별도의 긴 파이프 레일이 추가되어 레일 구조물이 중복 생성됨(고정 구조 변형/증식)",
       "인물의 하반신·다리가 선체 밖 공중에 완전히 떠 있고 물과의 접촉·접지점이 없어 상승을 만든 지지·반동이 존재하지 않음(발이 닿은 수면도, 밀어낸 선체면도 없음)"
      ],
      "physics": "한 손은 레일을 잡고 있으나, 다리 전체가 선체 우측 공중에 수평으로 뻗어 있고 물에 잠긴 부분이 없다. 물보라는 다리 뒤쪽에 흩뿌려져 있으나 발과 수면의 접촉점은 보이지 않는다. '몸의 반이 바닷물에 빠진 상태'라는 지문과 어긋나며, 도약을 만든 발판(수면 킥이나 선체 접촉)이 화면에 없다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "참조 순찰정 구조(짧은 선수 난간 2개, 선수 곡선·방현재)를 지키면서, 한 손은 선수 레일을 짚고 하반신은 여전히 물에 잠겨 물보라와 함께 솟구치는 지문의 '반신 침수 + 한손 지지 + 상승' 순간을 그대로 만든다. 인물 동일성·주간광·와이드 스케일도 충족."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "레일 라인을 매끄럽게 정리했지만, 참조에 없는 선수 감싸기 곡면 난간과 추가 파이프를 만들어 고정 구조를 변형·증식시켰고 선수 곡선부가 소실됐으며, 무엇보다 다리가 물에서 완전히 빠져 공중에 수평으로 뜬 채 지지·반동이 없어 '몸의 반이 바닷물에 빠진 상태'라는 지문을 깨뜨린다 — 얻은 것보다 잃은 것이 크다."
     }
    ],
    "all_candidates_fail": false
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "참조 순찰정 구조(짧은 선수 난간 2개, 선수 곡선·방현재)를 지키면서, 한 손은 선수 레일을 짚고 하반신은 여전히 물에 잠겨 물보라와 함께 솟구치는 지문의 '반신 침수 + 한손 지지 + 상승' 순간을 그대로 만든다. 인물 동일성·주간광·와이드 스케일도 충족."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "레일 라인을 매끄럽게 정리했지만, 참조에 없는 선수 감싸기 곡면 난간과 추가 파이프를 만들어 고정 구조를 변형·증식시켰고 선수 곡선부가 소실됐으며, 무엇보다 다리가 물에서 완전히 빠져 공중에 수평으로 뜬 채 지지·반동이 없어 '몸의 반이 바닷물에 빠진 상태'라는 지문을 깨뜨린다 — 얻은 것보다 잃은 것이 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "고산들의 시선과 상체는 프레임 좌하 전방(카메라 쪽)을 향하지만 몸통의 진행 방향은 레일을 넘어 갑판 쪽으로 솟구치는 방향으로 읽힌다. 왼손(화면 좌측 손)이 선수 레일 바를 실제로 움켜쥐고 있고, 반대 팔은 뒤로 뻗어 반동을 만든다. 뻗은 팔·머리·시선이 모두 갑판 상방으로 이어지는 상승선과 일치한다.",
      "built_space": "순찰정은 참조 사진의 구조와 일치한다 — 흰 조타실, 파란 갑판, 검은 선체, '한강경찰대'/'KOREA POLICE'/'한강경찰' 표기, 경찰 마크, 경광등, 조타실 상부 서치라이트 배열. 선수 좌현 난간이 별도로 서 있고, 인물이 잡은 레일은 선수 가장자리 바로 프레임 중앙을 대각으로 가로지른다. 레일 개수·배치가 참조와 큰 모순 없음. 다만 인물이 잡은 레일 하단부가 선체 상부 흰 판과 겹치며 접합이 다소 모호하다.",
      "entities": "인물은 40대 중반 한국 남성, 아주 짧게 깎은 검은 머리, 각진 차가운 얼굴 — 캐릭터 참조와 일치. 물에 흠뻑 젖은 어두운 군용/작업복 상하의로 야수적 인상을 유지. 추가 인물 없음. 텍스트·워터마크 없음. 낮의 자연광, 해면 반사 정상.",
      "hard_violations": [],
      "physics": "지지점이 명확하다: 한 손이 레일을 확실히 잡고, 하반신은 여전히 물 속/수면에 잠겨 있으며 발 주변에서 물보라가 튄다. 오른쪽 무릎·정강이가 선체 흰 면에 닿아 밀어올리는 접촉이 보인다. 물튀김이 다리 주변에 집중되어 솟구침의 반동과 일치. 공중에 떠 있는 부양체 없음."
     },
     {
      "label": "B",
      "direction": "시선·상체 방향은 A와 사실상 동일하게 좌하 전방. 손은 레일을 잡고 있으나 잡은 레일이 선수 좌현 난간에서 뽑혀 나온 듯 길게 휘어져 프레임 우측 아래로 이어지며, 그 레일 자체가 선체와 연결되지 않은 구간을 가진다 — 즉 그가 붙잡은 '레일'의 종착점이 불명확하다.",
      "built_space": "선박 구조는 참조와 대체로 일치하나 선수부가 A와 달리 재구성되었다: 좌현 난간이 선수 전체를 감싸는 곡면 파이프 난간으로 바뀌었고, 그 난간에서 다시 별개의 긴 파이프가 인물의 손 아래로 뻗어 우현 난간까지 이어지는 이중 레일 구조가 생겼다. 참조 사진의 선수 난간은 짧은 두 개의 독립 난간이며 선수를 감싸는 연속 곡선 레일은 없다 — 구조물이 증식·변형됨. 또한 선수 갑판 앞부분(검은 방현재/선수 곡선)이 잘려 나가 파란 갑판이 공중에 뜬 형태로 보인다.",
      "entities": "인물 동일 — 40대 중반, 짧은 머리, 각진 얼굴, 젖은 어두운 작업복. 참조와 일치. 추가 인물 없음. 선체 표기 '서울경찰청','한강경찰','한강경찰대' 가독. 텍스트 오염 없음.",
      "hard_violations": [
       "참조 구조에 없는 선수 감싸기 곡면 난간 + 별도의 긴 파이프 레일이 추가되어 레일 구조물이 중복 생성됨(고정 구조 변형/증식)",
       "인물의 하반신·다리가 선체 밖 공중에 완전히 떠 있고 물과의 접촉·접지점이 없어 상승을 만든 지지·반동이 존재하지 않음(발이 닿은 수면도, 밀어낸 선체면도 없음)"
      ],
      "physics": "한 손은 레일을 잡고 있으나, 다리 전체가 선체 우측 공중에 수평으로 뻗어 있고 물에 잠긴 부분이 없다. 물보라는 다리 뒤쪽에 흩뿌려져 있으나 발과 수면의 접촉점은 보이지 않는다. '몸의 반이 바닷물에 빠진 상태'라는 지문과 어긋나며, 도약을 만든 발판(수면 킥이나 선체 접촉)이 화면에 없다."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "인물의 시선은 카메라 좌하단 방향, 즉 자신이 올라서려는 갑판 쪽이 아니라 프레임 밖 정면을 향해 노려본다. 상체는 레일 너머 갑판 쪽으로 기울고 왼팔은 뒤로 뻗어 반동을 만든다. 오른손은 흰 레일 바를 잡고 있어 잡는 대상 자체는 계약대로 선수 레일이다. 다만 진행 방향(갑판 위로 솟구침)과 몸의 축이 갑판이 아니라 프레임 좌측 바깥으로 향해 상승 벡터가 다소 어긋난다.",
      "built_space": "순찰정은 참조 사진의 흰 조타실+파란 갑판+검은 선체 구성을 따르나, 선수 갑판이 비정상적으로 크고 평평한 원형 플랫폼처럼 확장되어 레일이 프레임 왼쪽 끝까지 길게 휘어 나간다. 레일이 인물의 손을 지난 뒤 오른쪽으로도 이어져 선체 밖 허공을 가로지르는 구간이 생겨 구조가 두 겹으로 읽힌다. 조타실의 'KOREA POLICE / 한강경찰' 표기와 문 위치는 참조와 유사. 뱃머리 형상이 참조의 뾰족한 선수와 달리 넓은 타원 데크로 변형됨.",
      "entities": "40대 중반 한국 남성, 짧게 친 검은 머리, 각진 차가운 얼굴 — 캐릭터 참조와 일치. 의상은 참조의 낚시복 대신 바닷물에 젖은 어두운 작업복/전투복 상하의로, 육혈귀 각성 상태의 상황(물에 빠진 상태)에 맞게 해석됨. 다른 인물은 없음. 화면 내 텍스트는 선체 도색 문자뿐.",
      "hard_violations": [
       "선수 갑판이 참조 구조의 뾰족한 뱃머리 대신 거대한 원형 플랫폼으로 변형되고, 레일이 선체 밖 허공까지 이어져 구조가 물리적으로 성립하지 않음",
       "인물의 하반신과 다리가 물 표면·선체 어디에도 접촉하지 않고 레일을 잡은 손 하나만으로 몸 전체가 수평에 가깝게 공중에 걸쳐 있어, 발이 물 밖 허공에 뜬 채 지지점이 없음"
      ],
      "physics": "한 손이 레일을 잡고 있으나, 다리와 하반신이 물 밖으로 완전히 나와 있고 발 아래 물보라와 물 표면 사이에 접점이 보이지 않는다. '몸의 반이 바닷물에 빠진 상태'라는 지시와 달리 몸 전체가 물 위 허공에 거의 수평으로 떠 있으며, 물에서 밀어낸 도약의 발판이 보이지 않는다. 물보라는 화려하지만 발끝이 어디서 튀어 올랐는지 지지가 불명확하다."
     },
     {
      "label": "B",
      "direction": "시선은 좌하단 정면을 노려보고 상체는 레일 너머 갑판 방향으로 크게 기울어 상승 벡터가 갑판 쪽으로 이어진다. 오른손은 선수 레일 바를 확실히 움켜쥐고 있고, 그 손이 피벗이 되어 몸이 레일 위로 크레스팅한다. 하반신과 다리는 레일 아래 선수 물보라 속에 잠겨 있어 '레일이 해수면과 갑판을 나눈다'는 지시가 성립한다.",
      "built_space": "참조 사진의 구조를 비교적 충실히 유지: 뾰족한 선수, 파란 갑판면, 검은 방현재 띠, 흰 조타실과 창, '한강경찰대'·'KOREA POLICE 한강경찰' 표기, 청색 경찰 엠블럼. 레일은 선수 좌현에서 대각선으로 프레임 중앙을 가로지르고 그 뒤에 갑판 상면이 착지 목표로 보인다. 선체 측면에 참조에 있는 서울경찰청 문자가 일부 인물/물보라에 가려지고, 선수 측면에 경찰 엠블럼이 하나 추가된 정도가 차이. 구조 전체는 한 척, 레일 한 벌로 개수 정상.",
      "entities": "40대 중반 한국 남성, 짧은 검은 머리, 각지고 차가운 얼굴 — 캐릭터 참조 얼굴/체형과 일치. 젖은 어두운 작업복 상하의로 바닷물에 잠겼던 상태를 표현. 추가 인물 없음. 텍스트는 선체 도색 문자에 한정.",
      "hard_violations": [],
      "physics": "레일을 잡은 오른손이 명확한 지지점이고, 하반신은 선수 아래 물속에 잠겨 있으며 무릎 부근에서 물이 터져 나오는 반동이 보인다. 발과 종아리가 수면 아래에 있어 '몸의 반이 물에 빠진 상태에서 솟구쳐 오르는' 중간 순간으로 읽히고, 도약의 출발점(물)과 도착점(갑판)이 모두 프레임 안에 있다. 공중에 근거 없이 떠 있는 요소 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "레일 그립은 있으나 하반신이 물 밖 허공에 수평으로 떠 지지점이 없고, 선수가 참조에 없는 거대한 원형 데크로 변형되며 레일이 선체 밖 허공까지 이어져 구조·물리 모두 무너짐 — '몸의 반이 물에 빠진 상태'라는 핵심 지시도 놓침."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "하반신이 물에 잠긴 채 한 손으로 선수 레일을 짚고 갑판 쪽으로 솟구치는 순간이 정확히 성립하고, 참조 순찰정의 뾰족한 선수·표기·엠블럼 구조를 유지하며 레일이 대각선으로 화면을 가로질러 해수면과 갑판을 나눈다; 선체 측면 문자 일부 가림은 사소한 차이."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "레일 그립은 있으나 하반신이 물 밖 허공에 수평으로 떠 지지점이 없고, 선수가 참조에 없는 거대한 원형 데크로 변형되며 레일이 선체 밖 허공까지 이어져 구조·물리 모두 무너짐 — '몸의 반이 물에 빠진 상태'라는 핵심 지시도 놓침."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "하반신이 물에 잠긴 채 한 손으로 선수 레일을 짚고 갑판 쪽으로 솟구치는 순간이 정확히 성립하고, 참조 순찰정의 뾰족한 선수·표기·엠블럼 구조를 유지하며 레일이 대각선으로 화면을 가로질러 해수면과 갑판을 나눈다; 선체 측면 문자 일부 가림은 사소한 차이."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "인물의 시선은 카메라 좌하단 방향, 즉 자신이 올라서려는 갑판 쪽이 아니라 프레임 밖 정면을 향해 노려본다. 상체는 레일 너머 갑판 쪽으로 기울고 왼팔은 뒤로 뻗어 반동을 만든다. 오른손은 흰 레일 바를 잡고 있어 잡는 대상 자체는 계약대로 선수 레일이다. 다만 진행 방향(갑판 위로 솟구침)과 몸의 축이 갑판이 아니라 프레임 좌측 바깥으로 향해 상승 벡터가 다소 어긋난다.",
      "built_space": "순찰정은 참조 사진의 흰 조타실+파란 갑판+검은 선체 구성을 따르나, 선수 갑판이 비정상적으로 크고 평평한 원형 플랫폼처럼 확장되어 레일이 프레임 왼쪽 끝까지 길게 휘어 나간다. 레일이 인물의 손을 지난 뒤 오른쪽으로도 이어져 선체 밖 허공을 가로지르는 구간이 생겨 구조가 두 겹으로 읽힌다. 조타실의 'KOREA POLICE / 한강경찰' 표기와 문 위치는 참조와 유사. 뱃머리 형상이 참조의 뾰족한 선수와 달리 넓은 타원 데크로 변형됨.",
      "entities": "40대 중반 한국 남성, 짧게 친 검은 머리, 각진 차가운 얼굴 — 캐릭터 참조와 일치. 의상은 참조의 낚시복 대신 바닷물에 젖은 어두운 작업복/전투복 상하의로, 육혈귀 각성 상태의 상황(물에 빠진 상태)에 맞게 해석됨. 다른 인물은 없음. 화면 내 텍스트는 선체 도색 문자뿐.",
      "hard_violations": [
       "선수 갑판이 참조 구조의 뾰족한 뱃머리 대신 거대한 원형 플랫폼으로 변형되고, 레일이 선체 밖 허공까지 이어져 구조가 물리적으로 성립하지 않음",
       "인물의 하반신과 다리가 물 표면·선체 어디에도 접촉하지 않고 레일을 잡은 손 하나만으로 몸 전체가 수평에 가깝게 공중에 걸쳐 있어, 발이 물 밖 허공에 뜬 채 지지점이 없음"
      ],
      "physics": "한 손이 레일을 잡고 있으나, 다리와 하반신이 물 밖으로 완전히 나와 있고 발 아래 물보라와 물 표면 사이에 접점이 보이지 않는다. '몸의 반이 바닷물에 빠진 상태'라는 지시와 달리 몸 전체가 물 위 허공에 거의 수평으로 떠 있으며, 물에서 밀어낸 도약의 발판이 보이지 않는다. 물보라는 화려하지만 발끝이 어디서 튀어 올랐는지 지지가 불명확하다."
     },
     {
      "label": "A",
      "direction": "시선은 좌하단 정면을 노려보고 상체는 레일 너머 갑판 방향으로 크게 기울어 상승 벡터가 갑판 쪽으로 이어진다. 오른손은 선수 레일 바를 확실히 움켜쥐고 있고, 그 손이 피벗이 되어 몸이 레일 위로 크레스팅한다. 하반신과 다리는 레일 아래 선수 물보라 속에 잠겨 있어 '레일이 해수면과 갑판을 나눈다'는 지시가 성립한다.",
      "built_space": "참조 사진의 구조를 비교적 충실히 유지: 뾰족한 선수, 파란 갑판면, 검은 방현재 띠, 흰 조타실과 창, '한강경찰대'·'KOREA POLICE 한강경찰' 표기, 청색 경찰 엠블럼. 레일은 선수 좌현에서 대각선으로 프레임 중앙을 가로지르고 그 뒤에 갑판 상면이 착지 목표로 보인다. 선체 측면에 참조에 있는 서울경찰청 문자가 일부 인물/물보라에 가려지고, 선수 측면에 경찰 엠블럼이 하나 추가된 정도가 차이. 구조 전체는 한 척, 레일 한 벌로 개수 정상.",
      "entities": "40대 중반 한국 남성, 짧은 검은 머리, 각지고 차가운 얼굴 — 캐릭터 참조 얼굴/체형과 일치. 젖은 어두운 작업복 상하의로 바닷물에 잠겼던 상태를 표현. 추가 인물 없음. 텍스트는 선체 도색 문자에 한정.",
      "hard_violations": [],
      "physics": "레일을 잡은 오른손이 명확한 지지점이고, 하반신은 선수 아래 물속에 잠겨 있으며 무릎 부근에서 물이 터져 나오는 반동이 보인다. 발과 종아리가 수면 아래에 있어 '몸의 반이 물에 빠진 상태에서 솟구쳐 오르는' 중간 순간으로 읽히고, 도약의 출발점(물)과 도착점(갑판)이 모두 프레임 안에 있다. 공중에 근거 없이 떠 있는 요소 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 8
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S87sh2__bgfirst_bg.png",
   "bg_asset_id": "d126da23-149d-4adc-b5c5-6a8420830ec1",
   "bg_record_key": "S87sh2::bgfirst_bg",
   "chain_winner": false,
   "authority": "seed_bg"
  },
  "ref_mode": "seed-bg+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  },
  "lane_policy": "ab_select_ready"
 },
 "groupbg::섬 갯바위": {
  "input_fingerprint": "f4a179920e3fbd9a",
  "meta": {
   "model": "gpt-image-2",
   "size": "1536x864",
   "pack": "11.202607220237",
   "contract": "bgfirst_full_v3",
   "group_sig": {
    "key": "섬 갯바위",
    "tags": [
     "S103sh2",
     "S103sh4",
     "S89sh3"
    ]
   },
   "context_sig": "e57154544bdbc343"
  },
  "prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated: The island's rocky shoreline beside the patrol boat's return route, where the villagers gather to receive it.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 갯바위: 거친 파도가 끊임없이 밀려와 부딪히는 검고 회색빛 도는 거대한 암초 표면. 물에 젖어 번들거리며 울퉁불퉁하고 날카로운 질감을 지닌 바위 지형. (특징: 거칠고 날카로운 암초 표면; 바위에 부서지는 거센 하얀 파도; 물에 젖어 번들거리는 바위 질감)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 갯바위 - 계속\n- 실외. 갯바위 - 저녁\n- 갯바위를 콱 움켜잡고 바다에서 나오는 사람은… 산들.\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "effective_prompt": "Create ONE empty live-action location background photograph — NO PEOPLE, no figures, no body parts, no silhouettes, no shadows or reflections of people anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods and produce in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch of a shot that happens at this location: use it ONLY as spatial evidence — what this place contains, how its ground, structures and landmarks are arranged and proportioned. Ignore the sketched people and arrows entirely, and do NOT copy its line style: render a fully photographic, physically plausible real place that fits THE LOCATION text below.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nTHE LOCATION — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated: The island's rocky shoreline beside the patrol boat's return route, where the villagers gather to receive it.\n\nLOCATION DETAIL (descriptive evidence about this place from the production's location records — evidence, not a staging order): use it to understand what kind of place this is and what it permanently contains. Carry over ONLY its enduring physical features — layout, structures, ground and materials, fixed equipment and installations. Do NOT treat any lighting, weather or time-of-day wording, momentary object states, depicted events or subjective impressions in it as fixed facts of the place; the TIME OF DAY lock below is the sole authority for time and lighting.\n금월도 갯바위: 거친 파도가 끊임없이 밀려와 부딪히는 검고 회색빛 도는 거대한 암초 표면. 물에 젖어 번들거리며 울퉁불퉁하고 날카로운 질감을 지닌 바위 지형. (특징: 거칠고 날카로운 암초 표면; 바위에 부서지는 거센 하얀 파도; 물에 젖어 번들거리는 바위 질감)\n\nSCENE EVIDENCE (verbatim quotes from the screenplay about this place — treat them as evidence of what the location physically contains and looks like; stage the PLACE those moments happen in, but do NOT depict the momentary actions, people or staged props themselves):\n- 실외. 금월도 갯바위 - 계속\n- 실외. 갯바위 - 저녁\n- 갯바위를 콱 움켜잡고 바다에서 나오는 사람은… 산들.\n\nTIME OF DAY (lock): day.\n\nRender ONE photorealistic empty location photograph, 16:9, neutral enough that every shot of this place can be staged from it later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_섬_갯바위_aa4183.png",
  "asset_id": "056f0676-e314-4bd5-b17f-9ff6683ae557",
  "input_asset_ids": [
   "8ff87c42-4044-4180-862d-909ea2703e54"
  ],
  "origin_tag": "S89sh3",
  "place_text": "The island's rocky shoreline beside the patrol boat's return route, where the villagers gather to receive it.",
  "origin_inputs": {
   "place_text": "The island's rocky shoreline beside the patrol boat's return route, where the villagers gather to receive it.",
   "time_of_day_en": "day",
   "conti_asset_id": "8ff87c42-4044-4180-862d-909ea2703e54"
  }
 },
 "S89sh3::bgfirst_bg": {
  "input_fingerprint": "f9bdc52eb5bdd211",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 선착장에 모인 채 배를 향해 팔을 뻗고 기괴하게 활짝 웃고 있는 금월도 마을 주민들의 전경.\n\nLOCATION (lock): The island's rocky shoreline beside the patrol boat's return route, where the villagers gather to receive it.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the eye-level dolly-out in a broad wide frame, increasing only the camera distance until the villagers read as a full collective tableau rather than isolated faces. Place the group across the middle of the frame with arms thrust toward the boat-facing camera line, their broad smiles and forward-reaching hands creating disturbing depth without exaggerated foreground scale.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 선착장 (Occupied by the gathered villagers); used as Supports the spread of the gathered villagers across the frame and fixes their relationship to the approaching boat direction; 바다 방향 (The patrol boat is approaching from this direction); used as Defines the off-camera direction toward which the villagers reach.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural ambient exterior light is treated with restrained contrast and subdued color so the villagers’ smiles and synchronized arms carry the unease.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 선착장에 모인 채 배를 향해 팔을 뻗고 기괴하게 활짝 웃고 있는 금월도 마을 주민들의 전경.\n\nLOCATION (lock): The island's rocky shoreline beside the patrol boat's return route, where the villagers gather to receive it.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the eye-level dolly-out in a broad wide frame, increasing only the camera distance until the villagers read as a full collective tableau rather than isolated faces. Place the group across the middle of the frame with arms thrust toward the boat-facing camera line, their broad smiles and forward-reaching hands creating disturbing depth without exaggerated foreground scale.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 선착장 (Occupied by the gathered villagers); used as Supports the spread of the gathered villagers across the frame and fixes their relationship to the approaching boat direction; 바다 방향 (The patrol boat is approaching from this direction); used as Defines the off-camera direction toward which the villagers reach.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural ambient exterior light is treated with restrained contrast and subdued color so the villagers’ smiles and synchronized arms carry the unease.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S89sh3__bgfirst_bg.png",
  "asset_id": "baf48e77-8523-4d51-9ca5-3ed096664a5f",
  "input_asset_ids": [
   "8ff87c42-4044-4180-862d-909ea2703e54",
   "056f0676-e314-4bd5-b17f-9ff6683ae557"
  ]
 },
 "S89sh3": {
  "input_fingerprint": "658befc11ecb4ab1",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 선착장에 모인 채 배를 향해 팔을 뻗고 기괴하게 활짝 웃고 있는 금월도 마을 주민들의 전경.\n\nLOCATION (lock): The island's rocky shoreline beside the patrol boat's return route, where the villagers gather to receive it. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the eye-level dolly-out in a broad wide frame, increasing only the camera distance until the villagers read as a full collective tableau rather than isolated faces. Place the group across the middle of the frame with arms thrust toward the boat-facing camera line, their broad smiles and forward-reaching hands creating disturbing depth without exaggerated foreground scale.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 선착장 (Occupied by the gathered villagers); used as Supports the spread of the gathered villagers across the frame and fixes their relationship to the approaching boat direction; 바다 방향 (The patrol boat is approaching from this direction); used as Defines the off-camera direction toward which the villagers reach.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural ambient exterior light is treated with restrained contrast and subdued color so the villagers’ smiles and synchronized arms carry the unease.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandel remains wet and wounded among the villagers, while Youngran retains her red shaman robe.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 마을 청년 #1 (Korean young man, short dark hair, broad youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 선착장에 모인 채 배를 향해 팔을 뻗고 기괴하게 활짝 웃고 있는 금월도 마을 주민들의 전경.\n\nLOCATION (lock): The island's rocky shoreline beside the patrol boat's return route, where the villagers gather to receive it. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the eye-level dolly-out in a broad wide frame, increasing only the camera distance until the villagers read as a full collective tableau rather than isolated faces. Place the group across the middle of the frame with arms thrust toward the boat-facing camera line, their broad smiles and forward-reaching hands creating disturbing depth without exaggerated foreground scale.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 선착장 (Occupied by the gathered villagers); used as Supports the spread of the gathered villagers across the frame and fixes their relationship to the approaching boat direction; 바다 방향 (The patrol boat is approaching from this direction); used as Defines the off-camera direction toward which the villagers reach.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural ambient exterior light is treated with restrained contrast and subdued color so the villagers’ smiles and synchronized arms carry the unease.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandel remains wet and wounded among the villagers, while Youngran retains her red shaman robe.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 마을 청년 #1 (Korean young man, short dark hair, broad youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 선착장에 모인 채 배를 향해 팔을 뻗고 기괴하게 활짝 웃고 있는 금월도 마을 주민들의 전경.\n\nLOCATION (lock): The island's rocky shoreline beside the patrol boat's return route, where the villagers gather to receive it. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Finish the eye-level dolly-out in a broad wide frame, increasing only the camera distance until the villagers read as a full collective tableau rather than isolated faces. Place the group across the middle of the frame with arms thrust toward the boat-facing camera line, their broad smiles and forward-reaching hands creating disturbing depth without exaggerated foreground scale.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 선착장 (Occupied by the gathered villagers); used as Supports the spread of the gathered villagers across the frame and fixes their relationship to the approaching boat direction; 바다 방향 (The patrol boat is approaching from this direction); used as Defines the off-camera direction toward which the villagers reach.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural ambient exterior light is treated with restrained contrast and subdued color so the villagers’ smiles and synchronized arms carry the unease.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandel remains wet and wounded among the villagers, while Youngran retains her red shaman robe.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 마을 청년 #1 (Korean young man, short dark hair, broad youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S89sh3__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S89sh3.png"
    },
    {
     "label": "CHARACTER REFERENCE — 마을 청년 #1: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:976931>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_섬_갯바위_aa4183.png"
    },
    {
     "label": "CHARACTER REFERENCE — 마을 청년 #1: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:976931>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "로케이션 재현과 물리적 지지는 양호하고 붉은 무복·젖은 넝마 인물로 CARRIED STATE를 반영했으나, 주민들이 선착장이 아닌 바위 사면에 모여 있고 중앙 인물 2~3명은 팔을 뻗지 않아 '배를 향해 일제히 팔을 뻗는' 집단 타블로의 통일성이 약하다.  ★위반: [gpt] 마을 청년 #1 이외의 남녀노소 주민 수십 명과 붉은 무복 여성을 추가해 명시된 인물 제한을 위반했다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "선착장 콘크리트 상판 위에 모인 주민 전원이 배(=카메라) 방향으로 팔을 뻗고 기괴하게 웃는 전경을 눈높이 와이드로 정확히 구현했고 로케이션 고정 요소도 일치 — 산델·영란의 CARRIED STATE 미반영이 유일한 감점.  ★위반: [gpt] 마을 청년 #1 이외의 중년 남녀 등 주민 수십 명을 추가해 명시된 인물 제한을 위반했다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.417,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.167,
      "B": 1.75
     },
     "violations": {
      "A": [
       "[gpt] 마을 청년 #1 이외의 남녀노소 주민 수십 명과 붉은 무복 여성을 추가해 명시된 인물 제한을 위반했다."
      ],
      "B": [
       "[gpt] 마을 청년 #1 이외의 중년 남녀 등 주민 수십 명을 추가해 명시된 인물 제한을 위반했다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "로케이션 재현과 물리적 지지는 양호하고 붉은 무복·젖은 넝마 인물로 CARRIED STATE를 반영했으나, 주민들이 선착장이 아닌 바위 사면에 모여 있고 중앙 인물 2~3명은 팔을 뻗지 않아 '배를 향해 일제히 팔을 뻗는' 집단 타블로의 통일성이 약하다.  ★위반: [gpt] 마을 청년 #1 이외의 남녀노소 주민 수십 명과 붉은 무복 여성을 추가해 명시된 인물 제한을 위반했다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "선착장 콘크리트 상판 위에 모인 주민 전원이 배(=카메라) 방향으로 팔을 뻗고 기괴하게 웃는 전경을 눈높이 와이드로 정확히 구현했고 로케이션 고정 요소도 일치 — 산델·영란의 CARRIED STATE 미반영이 유일한 감점.  ★위반: [gpt] 마을 청년 #1 이외의 중년 남녀 등 주민 수십 명을 추가해 명시된 인물 제한을 위반했다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "마을 주민 수십 명이 선착장 콘크리트 위에 도열해 모두 카메라 쪽(배가 접근하는 바다 방향 축)으로 두 팔을 뻗고 있다. 시선과 미소도 전원 렌즈를 향하며, 뻗은 손끝이 카메라 라인에 정확히 착지한다. 촬영은 눈높이, 배가 오는 방향에서 본 시점으로 읽힌다.",
      "built_space": "레퍼런스 장소의 콘크리트 선착장, 돌 축대, 흰 슬레이트 지붕 집 두 채, 전봇대, 오른쪽 검은 암반과 바다까지 구조가 대응한다. 부두 끝 계선주(검은 기둥)도 레퍼런스와 같은 위치에 보인다. 인물은 모두 선착장 노면 위에 서 있어 배치가 구조와 맞다.",
      "entities": "전원 한국인 마을 주민(성인 남녀 중심), 낡은 작업복·현대~근세 혼합 의상. 앞줄 중앙에 청년 #1과 유사한 짧은 검은 머리·넓은 얼굴의 청년이 보이나 두건은 없다. 붉은 무복의 영란, 젖은 부상자 산델은 프레임 안에서 식별되지 않는다(다만 SHOT TEXT는 주민 전경만 요구).",
      "physics": "모든 인물이 두 발로 노면을 딛고 상체를 앞으로 기울인 자세로, 무게 이동과 지지가 명확하다. 공중에 뜬 신체나 지지 없는 물체는 없다. 옷 주름, 젖은 바위, 파도 표면 모두 물리적 재질감이 유지된다.",
      "hard_violations": []
     },
     {
      "label": "B",
      "direction": "주민들이 대부분 왼쪽 프레임 밖 또는 화면 우측을 향해 옆으로 팔을 뻗고 있어, 뻗은 팔의 축이 카메라(배 방향)가 아니라 화면 내 수평 방향이다. 중앙에 선 여성과 붉은 옷 여성만 카메라를 본다. '카메라 라인을 향해 팔을 뻗는다'는 지시와 방향이 어긋난다.",
      "built_space": "같은 장소의 집·전봇대·축대·암반이 재현되었으나, 주민들이 선착장 콘크리트가 아니라 젖은 갯바위 위에 서 있다. 콘크리트 부두면은 좌측 끝에만 조금 보이고, 계선주는 오른쪽 먼 바위 위에 어색하게 놓여 있다.",
      "entities": "한국인 주민들이지만 아동이 다수 포함되고 근세 한복 계열 의상이 우세해 현대 어촌 주민보다 시대극 인상이다. 붉은 무복 여성(영란)이 등장하나 PEOPLE 조항은 마을 청년 #1만 허용한다. 중앙 인물은 청년 #1의 두건·누더기 의상을 입었으나 얼굴은 여성적으로 보여 캐릭터 일치도가 낮다.",
      "physics": "모든 인물이 바위 위에 발을 딛고 있고 공중 부양은 없다. 다만 다수가 미끄러운 젖은 바위 가장자리·물가에 서 있어 무게중심이 다소 불안하고, 팔을 옆으로 뻗은 동작의 지지는 유지된다.",
      "hard_violations": []
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "선착장 위에 모인 주민 전원이 배(카메라) 방향으로 팔을 뻗고 활짝 웃는 순간을 눈높이 광각 집단 전경으로 정확히 구현했고, 로케이션 구조와 자연광 톤도 레퍼런스에 부합한다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "장소 재현은 준수하나 주민들이 선착장이 아닌 갯바위에 서 있고 팔이 카메라(배) 방향이 아니라 측면으로 뻗어 있어 핵심 방향·무대 지시를 어겼으며, 아동·시대극 의상 등 지정되지 않은 인물 구성도 감점 요인이다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "선착장 위에 모인 주민 전원이 배(카메라) 방향으로 팔을 뻗고 활짝 웃는 순간을 눈높이 광각 집단 전경으로 정확히 구현했고, 로케이션 구조와 자연광 톤도 레퍼런스에 부합한다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "장소 재현은 준수하나 주민들이 선착장이 아닌 갯바위에 서 있고 팔이 카메라(배) 방향이 아니라 측면으로 뻗어 있어 핵심 방향·무대 지시를 어겼으며, 아동·시대극 의상 등 지정되지 않은 인물 구성도 감점 요인이다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "마을 주민 수십 명이 선착장 콘크리트 위에 도열해 모두 카메라 쪽(배가 접근하는 바다 방향 축)으로 두 팔을 뻗고 있다. 시선과 미소도 전원 렌즈를 향하며, 뻗은 손끝이 카메라 라인에 정확히 착지한다. 촬영은 눈높이, 배가 오는 방향에서 본 시점으로 읽힌다.",
      "built_space": "레퍼런스 장소의 콘크리트 선착장, 돌 축대, 흰 슬레이트 지붕 집 두 채, 전봇대, 오른쪽 검은 암반과 바다까지 구조가 대응한다. 부두 끝 계선주(검은 기둥)도 레퍼런스와 같은 위치에 보인다. 인물은 모두 선착장 노면 위에 서 있어 배치가 구조와 맞다.",
      "entities": "전원 한국인 마을 주민(성인 남녀 중심), 낡은 작업복·현대~근세 혼합 의상. 앞줄 중앙에 청년 #1과 유사한 짧은 검은 머리·넓은 얼굴의 청년이 보이나 두건은 없다. 붉은 무복의 영란, 젖은 부상자 산델은 프레임 안에서 식별되지 않는다(다만 SHOT TEXT는 주민 전경만 요구).",
      "physics": "모든 인물이 두 발로 노면을 딛고 상체를 앞으로 기울인 자세로, 무게 이동과 지지가 명확하다. 공중에 뜬 신체나 지지 없는 물체는 없다. 옷 주름, 젖은 바위, 파도 표면 모두 물리적 재질감이 유지된다.",
      "hard_violations": []
     },
     {
      "label": "A",
      "direction": "주민들이 대부분 왼쪽 프레임 밖 또는 화면 우측을 향해 옆으로 팔을 뻗고 있어, 뻗은 팔의 축이 카메라(배 방향)가 아니라 화면 내 수평 방향이다. 중앙에 선 여성과 붉은 옷 여성만 카메라를 본다. '카메라 라인을 향해 팔을 뻗는다'는 지시와 방향이 어긋난다.",
      "built_space": "같은 장소의 집·전봇대·축대·암반이 재현되었으나, 주민들이 선착장 콘크리트가 아니라 젖은 갯바위 위에 서 있다. 콘크리트 부두면은 좌측 끝에만 조금 보이고, 계선주는 오른쪽 먼 바위 위에 어색하게 놓여 있다.",
      "entities": "한국인 주민들이지만 아동이 다수 포함되고 근세 한복 계열 의상이 우세해 현대 어촌 주민보다 시대극 인상이다. 붉은 무복 여성(영란)이 등장하나 PEOPLE 조항은 마을 청년 #1만 허용한다. 중앙 인물은 청년 #1의 두건·누더기 의상을 입었으나 얼굴은 여성적으로 보여 캐릭터 일치도가 낮다.",
      "physics": "모든 인물이 바위 위에 발을 딛고 있고 공중 부양은 없다. 다만 다수가 미끄러운 젖은 바위 가장자리·물가에 서 있어 무게중심이 다소 불안하고, 팔을 옆으로 뻗은 동작의 지지는 유지된다.",
      "hard_violations": []
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 10,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 10,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "로케이션 재현과 물리적 지지는 양호하고 붉은 무복·젖은 넝마 인물로 CARRIED STATE를 반영했으나, 주민들이 선착장이 아닌 바위 사면에 모여 있고 중앙 인물 2~3명은 팔을 뻗지 않아 '배를 향해 일제히 팔을 뻗는' 집단 타블로의 통일성이 약하다.  ★위반: [gpt] 마을 청년 #1 이외의 남녀노소 주민 수십 명과 붉은 무복 여성을 추가해 명시된 인물 제한을 위반했다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "선착장 콘크리트 상판 위에 모인 주민 전원이 배(=카메라) 방향으로 팔을 뻗고 기괴하게 웃는 전경을 눈높이 와이드로 정확히 구현했고 로케이션 고정 요소도 일치 — 산델·영란의 CARRIED STATE 미반영이 유일한 감점.  ★위반: [gpt] 마을 청년 #1 이외의 중년 남녀 등 주민 수십 명을 추가해 명시된 인물 제한을 위반했다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/groupbg_섬_갯바위_aa4183.png"
   },
   {
    "label": "CHARACTER REFERENCE — 마을 청년 #1: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:976931>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "군중이 현대 복장을 한 다양한 인물들로 구성되어 있어 '마을 청년 #1'만 등장해야 한다는 지시를 어겼으며, 지정된 붉은 무당복을 입은 영란과 젖고 다친 산델이 군중 속에 없습니다.",
     "fix_en": "Replace all individuals in the crowd with the specific young man from the character reference, wearing his ragged Joseon clothing. Change one front-row figure's clothing to a bright red shaman robe, and render another front-row figure as soaking wet with visible wounds. Preserve the current camera framing, the lighting, the rocky shoreline, the pier, the houses, and the exact positions and reaching poses of the crowd.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace all individuals in the crowd with the specific young man from the character reference, wearing his ragged Joseon clothing. Change one front-row figure's clothing to a bright red shaman robe, and render another front-row figure as soaking wet with visible wounds. Preserve the current camera framing, the lighting, the rocky shoreline, the pier, the houses, and the exact positions and reaching poses of the crowd.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "군중 전원이 카메라(=배가 접근하는 바다 방향) 쪽으로 팔을 뻗고 시선도 렌즈 축을 향한다. 앞줄부터 부두 끝까지 손바닥이 카메라 라인에 정확히 겨눠져 있고, 뒷줄 일부는 팔을 위로 들어 흔든다. 목표는 오프카메라의 순찰선 방향으로 일치.",
      "built_space": "레퍼런스의 콘크리트 선착장, 돌 축대, 기와 얹은 흰 벽 가옥 2채, 전신주와 전선, 오른쪽 검은 갯바위와 계선주(볼라드)까지 위치와 개수가 로케이션 사진과 대응한다. 군중은 선착장 위에만 서 있어 지면 관계가 맞다.",
      "entities": "40여 명의 남녀노소 마을 주민 — 중년 남녀, 노인, 청년이 섞여 있어 '마을 주민들'의 집합으로 읽힌다. 다만 컨트랙트가 지정한 '마을 청년 #1' 캐릭터 레퍼런스(수건 두건, 누더기 한복)와 일치하는 인물은 뚜렷하지 않고 의상은 현대 시골 작업복 계열. 붉은 무복의 영란, 젖고 부상한 산델은 프레임 안에 식별되지 않음.",
      "hard_violations": [],
      "physics": "모든 인물이 두 발로 콘크리트 노면을 딛고 있으며 상체를 앞으로 기울인 만큼 무게중심이 앞발에 실려 있다. 공중에 뜬 신체나 지지 없는 물체 없음. 그림자도 발밑에 붙어 있다."
     },
     {
      "label": "B",
      "direction": "역시 전원이 카메라 축(배 접근 방향)으로 팔을 뻗고 응시한다. 조준 방향 자체는 정확하나 표정이 '기괴하게 활짝 웃는' 것이 아니라 무표정·경직에 가깝다.",
      "built_space": "A와 동일한 로케이션 지오메트리(선착장, 축대, 가옥, 전신주, 볼라드, 갯바위)가 유지된다. 배경 처리는 문제 없음.",
      "entities": "군중이 거의 전원 10~20대 남성으로 치환되었고 다수가 서로 매우 닮은 복제 같은 얼굴이다. 노인·여성·아이가 사라져 '마을 주민들'이 아니라 청년 집단으로 읽힌다. 앞줄에 붉은 한복(영란 추정) 인물이 하나 추가되어 있으나 컨트랙트의 PEOPLE 조항은 이 샷에 마을 청년 #1만을 허용한다.",
      "hard_violations": [
       "동일 인물로 보이는 얼굴이 군중 전반에 반복 복제됨(중복 신체)",
       "샷 텍스트가 부르지 않은 붉은 무복 인물을 전경에 추가 배치(발명된 인물)"
      ],
      "physics": "인물들은 지면을 딛고 있고 부유체는 없다. 다만 다수가 동일한 각도로 굳어 있어 체중 이동이 거의 읽히지 않는다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "남녀노소가 섞인 실제 '마을 주민' 집합이 선착장 전체에 퍼져 카메라(배) 방향으로 팔을 뻗고 기괴하게 활짝 웃는 와이드 전경 — 샷 텍스트의 웃음·집단 타블로·로케이션 락을 모두 충족하고 복제 얼굴이나 추가 인물이 없다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "조준 방향과 로케이션은 맞지만 주민 전원을 닮은꼴 청년 남성으로 바꿔 얼굴이 복제되고, 샷이 부르지 않은 붉은 무복 인물을 추가했으며 '기괴하게 활짝 웃는' 핵심 연기가 무표정으로 사라져 계약이 요구하지 않은 손실이 크다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "남녀노소가 섞인 실제 '마을 주민' 집합이 선착장 전체에 퍼져 카메라(배) 방향으로 팔을 뻗고 기괴하게 활짝 웃는 와이드 전경 — 샷 텍스트의 웃음·집단 타블로·로케이션 락을 모두 충족하고 복제 얼굴이나 추가 인물이 없다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "조준 방향과 로케이션은 맞지만 주민 전원을 닮은꼴 청년 남성으로 바꿔 얼굴이 복제되고, 샷이 부르지 않은 붉은 무복 인물을 추가했으며 '기괴하게 활짝 웃는' 핵심 연기가 무표정으로 사라져 계약이 요구하지 않은 손실이 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "군중 전원이 카메라(=배가 접근하는 바다 방향) 쪽으로 팔을 뻗고 시선도 렌즈 축을 향한다. 앞줄부터 부두 끝까지 손바닥이 카메라 라인에 정확히 겨눠져 있고, 뒷줄 일부는 팔을 위로 들어 흔든다. 목표는 오프카메라의 순찰선 방향으로 일치.",
      "built_space": "레퍼런스의 콘크리트 선착장, 돌 축대, 기와 얹은 흰 벽 가옥 2채, 전신주와 전선, 오른쪽 검은 갯바위와 계선주(볼라드)까지 위치와 개수가 로케이션 사진과 대응한다. 군중은 선착장 위에만 서 있어 지면 관계가 맞다.",
      "entities": "40여 명의 남녀노소 마을 주민 — 중년 남녀, 노인, 청년이 섞여 있어 '마을 주민들'의 집합으로 읽힌다. 다만 컨트랙트가 지정한 '마을 청년 #1' 캐릭터 레퍼런스(수건 두건, 누더기 한복)와 일치하는 인물은 뚜렷하지 않고 의상은 현대 시골 작업복 계열. 붉은 무복의 영란, 젖고 부상한 산델은 프레임 안에 식별되지 않음.",
      "hard_violations": [],
      "physics": "모든 인물이 두 발로 콘크리트 노면을 딛고 있으며 상체를 앞으로 기울인 만큼 무게중심이 앞발에 실려 있다. 공중에 뜬 신체나 지지 없는 물체 없음. 그림자도 발밑에 붙어 있다."
     },
     {
      "label": "B",
      "direction": "역시 전원이 카메라 축(배 접근 방향)으로 팔을 뻗고 응시한다. 조준 방향 자체는 정확하나 표정이 '기괴하게 활짝 웃는' 것이 아니라 무표정·경직에 가깝다.",
      "built_space": "A와 동일한 로케이션 지오메트리(선착장, 축대, 가옥, 전신주, 볼라드, 갯바위)가 유지된다. 배경 처리는 문제 없음.",
      "entities": "군중이 거의 전원 10~20대 남성으로 치환되었고 다수가 서로 매우 닮은 복제 같은 얼굴이다. 노인·여성·아이가 사라져 '마을 주민들'이 아니라 청년 집단으로 읽힌다. 앞줄에 붉은 한복(영란 추정) 인물이 하나 추가되어 있으나 컨트랙트의 PEOPLE 조항은 이 샷에 마을 청년 #1만을 허용한다.",
      "hard_violations": [
       "동일 인물로 보이는 얼굴이 군중 전반에 반복 복제됨(중복 신체)",
       "샷 텍스트가 부르지 않은 붉은 무복 인물을 전경에 추가 배치(발명된 인물)"
      ],
      "physics": "인물들은 지면을 딛고 있고 부유체는 없다. 다만 다수가 동일한 각도로 굳어 있어 체중 이동이 거의 읽히지 않는다."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "readings": [
     {
      "label": "A",
      "direction": "모든 주민이 카메라(=배가 접근하는 바다 방향) 쪽으로 두 팔을 뻗고 시선도 렌즈 축을 향한다. 팔 끝과 손바닥이 카메라 라인에 정확히 겨냥돼 있어 '배를 향해 팔을 뻗는' 방향은 성립. 다만 표정은 웃음이 거의 없고 무표정·긴장된 얼굴이 대부분이라 '기괴하게 활짝 웃고 있는'이라는 핵심 연기 지시가 성립하지 않는다.",
      "built_space": "레퍼런스의 콘크리트 선착장, 돌축대, 흰 슬레이트 지붕 가옥 2채, 전신주와 전선, 오른쪽 계선주(볼라드), 검은 암반과 바다까지 위치·비율이 잘 맞는다. 군중은 선착장 위에 한 줄로 길게 늘어서 원근에 따라 축소되며, 배경 가옥은 실제 크기대로 작게 남아 있다.",
      "entities": "거의 모든 인물이 20대 초반 남성으로, 캐릭터 레퍼런스의 청년 얼굴이 수십 번 복제된 듯 동일한 얼굴/헤어가 반복된다. 마을 '주민들'이라는 집단(노소·남녀 혼재)이 아니라 동일 인물 클론 군집으로 읽혀 복제 신체 문제가 발생한다. 붉은 무복의 영란은 프레임 좌중앙에 있으나 얼굴이 주변 청년들과 비슷한 앳된 남성형에 가깝다. 젖고 부상한 산델로 특정할 인물은 불분명.",
      "hard_violations": [
       "동일 인물(마을 청년 #1)의 얼굴이 수십 명에게 그대로 복제되어 사실상 중복 신체 군집으로 읽힘"
      ],
      "physics": "전원이 두 발로 선착장 콘크리트를 딛고 상체를 앞으로 기울인 자세로, 체중 이동과 발 접지가 확인된다. 공중에 뜬 신체나 지지 없는 물체는 없다. 옷의 주름과 젖은 자국, 맨발의 흙때 등 재질감은 사실적."
     },
     {
      "label": "B",
      "direction": "주민들이 카메라(바다·배 방향)로 팔을 뻗고 손가락을 갈퀴처럼 벌린 채 렌즈를 향해 활짝 웃는다. 시선·손끝 모두 카메라 축에 겨냥돼 있어 샷 텍스트의 '배를 향해 팔을 뻗고 기괴하게 활짝 웃는' 지시가 방향과 표정 모두에서 충족된다.",
      "built_space": "동일한 선착장 콘크리트 슬래브, 돌축대, 흰 가옥 2채, 전신주, 오른쪽 볼라드, 검은 암반, 바다 지평선이 A와 거의 동일한 배치로 유지된다. 군중이 부두를 따라 원근으로 줄어들며 배경 요소의 실제 스케일도 지켜짐.",
      "entities": "남녀노소가 섞인 실제 마을 주민 집단(중년 여성, 노년 남성, 청년 등)으로 다양성이 확보돼 '마을 주민들'에 부합. 다만 계약의 PEOPLE 조항은 등장 인물을 '마을 청년 #1'로 한정하는데, 화면 앞줄은 대부분 중장년이고 청년 레퍼런스 얼굴은 중앙 우측 한 명 정도로만 근사한다. 붉은 무복(영란)은 보이지 않으며, 젖고 부상한 산델도 식별되지 않아 CARRIED STATE 조항이 이행되지 않았다.",
      "hard_violations": [],
      "physics": "전원이 부두 위에 두 발로 서서 상체를 앞으로 숙인 자세이며 접지와 무게중심이 자연스럽다. 공중에 뜬 신체나 지지 없는 물체 없음. 의복 주름, 흙 묻은 작업복, 젖은 콘크리트 반사 등 물성 표현이 사실적."
     }
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "'기괴하게 활짝 웃는' 핵심 연기와 팔 뻗는 방향을 모두 충족하고, 로케이션 지오메트리를 A와 동일하게 유지하면서 클론 복제 얼굴 문제를 해소했다. 대신 붉은 무복(영란)과 젖은 산델이 사라진 것이 대가지만, 얼굴 복제라는 치명적 결함 제거와 표정 지시 충족이 그 손실을 상회한다."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "프레이밍·로케이션·팔 방향은 맞고 붉은 무복 인물은 남아 있으나, 수십 명 전원이 캐릭터 레퍼런스 청년의 복제 얼굴로 채워져 중복 신체에 해당하며 '기괴하게 활짝 웃는' 표정 지시도 무표정으로 어겼다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "'기괴하게 활짝 웃는' 핵심 연기와 팔 뻗는 방향을 모두 충족하고, 로케이션 지오메트리를 A와 동일하게 유지하면서 클론 복제 얼굴 문제를 해소했다. 대신 붉은 무복(영란)과 젖은 산델이 사라진 것이 대가지만, 얼굴 복제라는 치명적 결함 제거와 표정 지시 충족이 그 손실을 상회한다."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "프레이밍·로케이션·팔 방향은 맞고 붉은 무복 인물은 남아 있으나, 수십 명 전원이 캐릭터 레퍼런스 청년의 복제 얼굴로 채워져 중복 신체에 해당하며 '기괴하게 활짝 웃는' 표정 지시도 무표정으로 어겼다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "모든 주민이 카메라(=배가 접근하는 바다 방향) 쪽으로 두 팔을 뻗고 시선도 렌즈 축을 향한다. 팔 끝과 손바닥이 카메라 라인에 정확히 겨냥돼 있어 '배를 향해 팔을 뻗는' 방향은 성립. 다만 표정은 웃음이 거의 없고 무표정·긴장된 얼굴이 대부분이라 '기괴하게 활짝 웃고 있는'이라는 핵심 연기 지시가 성립하지 않는다.",
      "built_space": "레퍼런스의 콘크리트 선착장, 돌축대, 흰 슬레이트 지붕 가옥 2채, 전신주와 전선, 오른쪽 계선주(볼라드), 검은 암반과 바다까지 위치·비율이 잘 맞는다. 군중은 선착장 위에 한 줄로 길게 늘어서 원근에 따라 축소되며, 배경 가옥은 실제 크기대로 작게 남아 있다.",
      "entities": "거의 모든 인물이 20대 초반 남성으로, 캐릭터 레퍼런스의 청년 얼굴이 수십 번 복제된 듯 동일한 얼굴/헤어가 반복된다. 마을 '주민들'이라는 집단(노소·남녀 혼재)이 아니라 동일 인물 클론 군집으로 읽혀 복제 신체 문제가 발생한다. 붉은 무복의 영란은 프레임 좌중앙에 있으나 얼굴이 주변 청년들과 비슷한 앳된 남성형에 가깝다. 젖고 부상한 산델로 특정할 인물은 불분명.",
      "hard_violations": [
       "동일 인물(마을 청년 #1)의 얼굴이 수십 명에게 그대로 복제되어 사실상 중복 신체 군집으로 읽힘"
      ],
      "physics": "전원이 두 발로 선착장 콘크리트를 딛고 상체를 앞으로 기울인 자세로, 체중 이동과 발 접지가 확인된다. 공중에 뜬 신체나 지지 없는 물체는 없다. 옷의 주름과 젖은 자국, 맨발의 흙때 등 재질감은 사실적."
     },
     {
      "label": "A",
      "direction": "주민들이 카메라(바다·배 방향)로 팔을 뻗고 손가락을 갈퀴처럼 벌린 채 렌즈를 향해 활짝 웃는다. 시선·손끝 모두 카메라 축에 겨냥돼 있어 샷 텍스트의 '배를 향해 팔을 뻗고 기괴하게 활짝 웃는' 지시가 방향과 표정 모두에서 충족된다.",
      "built_space": "동일한 선착장 콘크리트 슬래브, 돌축대, 흰 가옥 2채, 전신주, 오른쪽 볼라드, 검은 암반, 바다 지평선이 A와 거의 동일한 배치로 유지된다. 군중이 부두를 따라 원근으로 줄어들며 배경 요소의 실제 스케일도 지켜짐.",
      "entities": "남녀노소가 섞인 실제 마을 주민 집단(중년 여성, 노년 남성, 청년 등)으로 다양성이 확보돼 '마을 주민들'에 부합. 다만 계약의 PEOPLE 조항은 등장 인물을 '마을 청년 #1'로 한정하는데, 화면 앞줄은 대부분 중장년이고 청년 레퍼런스 얼굴은 중앙 우측 한 명 정도로만 근사한다. 붉은 무복(영란)은 보이지 않으며, 젖고 부상한 산델도 식별되지 않아 CARRIED STATE 조항이 이행되지 않았다.",
      "hard_violations": [],
      "physics": "전원이 부두 위에 두 발로 서서 상체를 앞으로 숙인 자세이며 접지와 무게중심이 자연스럽다. 공중에 뜬 신체나 지지 없는 물체 없음. 의복 주름, 흙 묻은 작업복, 젖은 콘크리트 반사 등 물성 표현이 사실적."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 7
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S89sh3__bgfirst_bg.png",
   "bg_asset_id": "baf48e77-8523-4d51-9ca5-3ed096664a5f",
   "bg_record_key": "S89sh3::bgfirst_bg",
   "chain_winner": false,
   "authority": "groupbg",
   "group_key": "섬 갯바위",
   "groupbg_asset_id": "056f0676-e314-4bd5-b17f-9ff6683ae557"
  },
  "ref_mode": "그룹 배경+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S93sh2": {
  "input_fingerprint": "759b681089e30dc7",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): dusk, golden sunset.\n\nSHOT TEXT (authoritative, Korean): 무언가를 품에 안은 듯한 낯선 실루엣으로 웅크려 앉아 있는 강민숙(홍은미)의 뒷모습.\n\nLOCATION (lock): The far stone end of the island pier, facing the gold-lit sea at sunset. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: End the eye-height dolly-in several meters behind the seated figure, tightening to a medium rear silhouette while keeping the cradled person mostly buried beneath the figure’s shoulders and arms. The seated figure stays centered and hunched toward the sea; only an incomplete, limp portion of the concealed body is allowed into the lower frame so neither face nor identity is disclosed.\n- FRAMING SCALE: medium shot\n- FRAME LAYOUT: identity-concealed seated figure in the middle-center of the frame, midground; mostly concealed person in her hold in the lower-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 선착장 돌바닥 (The figures are positioned near its end); used as Provides the straight approach axis and supports the lower edge of the hunched silhouette; 금월도 앞바다 (Reflecting golden sunset light); used as Forms the distant direction faced by the seated figure and supplies separation around the silhouette.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Golden sunset reflection outlines the hunched rear silhouette while the concealed body remains in heavier contrast.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): Her soaked corpse is held limp across 강영란(흰 머리 여자)’s lap and arms at the end of the dock, with her upper body and face concealed behind 강영란(흰 머리 여자)’s shoulder. Her head and limbs have no muscular support while the body is directed toward the sea.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 강민숙(홍은미) (Korean woman in her late 40s, dark hair, refined narrow face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "앉은 인물이 카메라 쪽(정면)을 향해 고개를 숙이고 있어 '뒷모습' 지시와 정반대다. 시선은 품에 안은 천 뭉치로 향하고, 바다는 인물의 등 뒤에 있다. 즉 바다를 향해 웅크린 뒷모습이 아니라 얼굴이 드러난 정면 쭈그려 앉은 자세.",
    "built_space": "선착장 돌바닥, 가로등, 어선, 어구 상자, 우측 처마 등 레퍼런스 구조물은 대체로 재현되나 카메라가 인물 앞(바다 쪽)에 서 있어 지시된 '수 미터 뒤' 카메라 위치와 불일치.",
    "entities": "인물은 40대 후반 한국 여성으로 보이나 얼굴이 완전히 노출되어 '얼굴·정체 비공개' 조항 위반. 품에 안긴 것은 어두운 천에 싸인 작은 뭉치로, 성인 여성의 시신(사지가 늘어진 몸)으로 읽히지 않고 아기 포대기처럼 보인다 — 즉 지정된 강민숙의 시신이 아님.",
    "hard_violations": [
     "샷 텍스트가 명시한 '뒷모습'을 정면 얼굴 노출로 렌더 — 정체 비공개 조항과 카메라 축 위반",
     "품에 안긴 대상이 성인 시신이 아닌 천에 싸인 작은 덩어리(아기 형상)로 대체된 발명된 오브젝트"
    ],
    "physics": "쭈그려 앉은 자세는 두 발바닥이 바닥에 닿아 지지되어 물리적으로 성립. 다만 안긴 덩어리는 팔로 들어올려 가슴에 붙여 안겨 있어, '무릎과 팔 위에 축 늘어져 지지된 시신' 조항과 어긋난다."
   },
   {
    "label": "B",
    "direction": "인물은 카메라에 등을 보이며 바다·석양 반사광 쪽을 정확히 향한다. 방향축은 지시대로 맞다. 다만 안긴 인물의 방향은 불분명하고 다리만 양옆으로 삐져나온다.",
    "built_space": "선착장 돌바닥과 바다만 보이고 가로등, 어선, 우측 건물 처마, 어구 등 레퍼런스의 고정 요소가 전무하다. 부두 끝이 매우 짧게 잘려 장소 잠금(로케이션 락) 재현이 약하다.",
    "entities": "실루엣이 짙어 성별·연령·머리색 판별이 어렵고, 상의도 레퍼런스의 누비 패딩이 아닌 밝은 셔츠로 읽힌다. 안긴 인물은 다리 두 쌍이 양옆으로 나와 형태가 뒤엉켜 있어 신체 구성이 모호하다.",
    "hard_violations": [
     "안긴 신체의 해부 구조가 불분명 — 좌우로 뻗은 다리/팔 형태가 앉은 인물의 신체와 뒤섞여 판독 불가한 덩어리를 형성"
    ],
    "physics": "앉은 인물은 바닥에 엉덩이와 발이 닿아 지지됨. 안긴 신체는 무릎 위에 걸쳐진 듯하나 다리 끝이 바닥에 닿아 있어 중력에는 대체로 부합. 다만 형태 판별이 어려워 지지점 확인이 제한적."
   },
   {
    "label": "C",
    "direction": "앉은 여성은 카메라에 등을 보이고 석양의 바다를 정면으로 바라본다 — 지시된 뒷모습·바다 방향과 정확히 일치. 안긴 시신의 머리는 여성 어깨 너머 오른쪽으로 늘어져 얼굴이 가려진다.",
    "built_space": "레퍼런스의 돌 포장 부두, 좌측 계류 밧줄·소형 선박, 우측 어구 상자와 통, 가로등 기둥, 원경의 섬 실루엣이 모두 유지되어 장소 잠금이 가장 충실. 카메라는 인물 수 미터 뒤 눈높이.",
    "entities": "앉은 인물은 어두운 묶은 머리, 레퍼런스와 동일한 베이지 누비 재킷 — 이전 컷 인물과 일치. 품에 안긴 인물은 젖은 검은 머리와 늘어진 팔·손이 보이는 성인 여성으로, 얼굴은 비공개. 다만 앉은 인물의 맨다리/맨발은 이전 컷의 바지 차림과 다소 어긋난다.",
    "hard_violations": [],
    "physics": "앉은 인물은 엉덩이와 다리로 돌바닥에 완전히 지지. 시신은 무릎·팔 위에 가로놓여 머리와 팔이 근력 없이 아래로 축 늘어지고 손등이 돌바닥에 닿아 있어 중력 순응 및 지지점이 명확하다."
   }
  ],
  "totals": {
   "A": 2,
   "B": 5,
   "C": 8
  },
  "selected": "C",
  "ranking": [
   "C",
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 2,
    "verdict_ko": "'뒷모습' 지시를 정면 얼굴 노출로 뒤집었고, 품에 안긴 것도 성인 시신이 아닌 천에 싸인 아기 형태라 정체 비공개·불동 포즈 계약을 모두 위반."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "뒷모습·바다 방향은 맞지만 레퍼런스의 부두 고정물(가로등·어선·어구·건물)이 전부 사라져 로케이션 락이 무너졌고, 안긴 신체가 판독 불가한 덩어리로 뭉개졌다."
   },
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "수 미터 뒤 눈높이 미디엄 뒷모습, 중앙 배치, 어깨 아래로 축 늘어진 시신의 일부만 노출, 레퍼런스 장소·의상 유지까지 모두 충족한 유일한 컷."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S92sh3_sel.png"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "안겨 있는 사람의 맨다리가 왼쪽에 길게 삐져나오고 오른쪽에는 얼굴이 그대로 노출되어, 얼굴과 정체를 숨기고 인물의 어깨 아래에 파묻히게 하라는 지시를 위반했습니다.",
     "fix_en": "Erase the bare legs protruding on the left and the visible head and face protruding on the right of the seated figure, replacing those specific areas with the stone pier and sea background. Completely envelop the cradled body behind the seated woman's back and dark quilted jacket, leaving only a small, unidentifiable shadowed lump in her lower-center lap. Preserve the seated woman's exact posture, dark hair, quilted jacket, the stone pier, the sunset lighting, and the overall framing exactly as they are.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 2,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Erase the bare legs protruding on the left and the visible head and face protruding on the right of the seated figure, replacing those specific areas with the stone pier and sea background. Completely envelop the cradled body behind the seated woman's back and dark quilted jacket, leaving only a small, unidentifiable shadowed lump in her lower-center lap. Preserve the seated woman's exact posture, dark hair, quilted jacket, the stone pier, the sunset lighting, and the overall framing exactly as they are.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "앉은 여성의 뒷모습이 화면 중앙에 있고, 몸통과 시선이 정면의 금빛 바다(석양 반사)를 향한다. 어깨선이 앞으로 숙여져 있어 품에 안은 대상 쪽으로 기울어져 있고, 카메라 축은 선착장 돌바닥의 직선 접근축을 따라 눈높이에서 뒤에서 바라본다. 안긴 인물의 머리는 앉은 여성의 오른쪽 어깨 너머 아래로 늘어져 있어 얼굴이 보이지 않는다.",
      "built_space": "선착장 돌바닥이 하단을 받치고, 좌측에 계류 로프·작은 배·난간 기둥, 우측에 통발 바구니 더미와 잡동사니, 화면 우상단으로 가로등 기둥 하나가 서 있다. 이전 스틸의 부두 끝·조명 기둥·통발 배치와 일치. 구조물 중복이나 잘못된 개수는 없음.",
      "entities": "앉은 인물은 40대 후반으로 보이는 한국 여성, 어두운 머리를 낮게 묶은 포니테일, 이전 스틸과 동일한 누비 카키 재킷·어두운 하의. 품에 안긴 인물은 젖은 듯 어두운 머리카락과 어두운 상의의 팔·손이 하단 우측으로 늘어져 부분만 보여, 신원 비노출 조건을 충족. 다만 안는 이의 왼쪽 종아리·맨발이 노출되어 이전 스틸의 복장(긴 바지)과 다소 어긋나며, 젖은 시신의 팔로도 읽힐 여지가 있는 애매한 다리 요소가 있다.",
      "hard_violations": [],
      "physics": "안긴 인물의 머리는 안는 이의 어깨/팔 위에 얹혀 중력대로 떨궈져 있고, 팔과 손은 돌바닥 쪽으로 완전히 늘어져 손등이 바닥에 닿아 있다. 근육 지지 없이 무릎과 팔로 받쳐지는 형태로, 지지점이 명확하다. 공중에 떠 있는 부분 없음. 다만 노출된 맨다리/발 하나가 누구 것인지 다소 모호한 해부학적 혼선이 있다."
     },
     {
      "label": "B",
      "direction": "동일한 뒷모습 앉은 자세, 몸과 시선은 금빛 바다를 향한다. 그러나 품에 안은 대상이 전혀 없고, 팔은 무릎 위에 얹혀 결가부좌에 가깝게 앉아 있다. 시선·자세가 향하는 목표는 바다뿐이며, 샷 텍스트가 요구한 '품에 안은 듯한 낯선 실루엣'의 대상이 화면에 존재하지 않는다.",
      "built_space": "선착장 돌바닥, 좌측 계류 로프·소형 배·난간, 우측 통발 더미, 우상단 기둥 — 이전 스틸과 재질·배치 모두 일치. 구조 오류 없음.",
      "entities": "앉은 인물은 40대 후반 한국 여성, 낮게 묶은 포니테일, 누비 카키 재킷·어두운 바지로 이전 스틸과 정확히 일치. 그러나 계약이 요구한 '안긴 채 은폐된 인물(강민숙의 시신)'이 아예 부재해 필수 엔티티 하나가 빠져 있다.",
      "hard_violations": [
       "샷 텍스트·프레임 레이아웃이 명시한 '품에 안긴 채 은폐된 인물(하단-중앙 미드그라운드)'이 전혀 렌더링되지 않음 — 필수 인물 누락"
      ],
      "physics": "여성은 돌바닥에 엉덩이와 다리로 완전히 앉아 있어 지지 문제 없음. 공중에 뜬 요소 없음. 다만 지지해야 할 시신 자체가 없어 물리 검증 대상이 성립하지 않는다."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "샷 텍스트 핵심인 '품에 안은 듯한 실루엣'과 어깨 너머로 얼굴이 가려진 채 중력대로 늘어진 시신의 머리·팔을 정확히 구현했고, 장소·의상·석양 톤도 이전 스틸과 일치 — 노출된 맨다리/발의 소속이 다소 모호하고 하의가 이전 스틸과 어긋나는 손실은 있으나 계약의 최상위 요구를 유일하게 충족한다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "장소·의상·광선은 이전 스틸과 깔끔히 일치하지만, 계약이 명시한 '품에 안긴 채 은폐된 인물'이 통째로 사라져 샷 텍스트와 프레임 레이아웃의 핵심 요소를 충족하지 못한 실패작이다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "샷 텍스트 핵심인 '품에 안은 듯한 실루엣'과 어깨 너머로 얼굴이 가려진 채 중력대로 늘어진 시신의 머리·팔을 정확히 구현했고, 장소·의상·석양 톤도 이전 스틸과 일치 — 노출된 맨다리/발의 소속이 다소 모호하고 하의가 이전 스틸과 어긋나는 손실은 있으나 계약의 최상위 요구를 유일하게 충족한다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "장소·의상·광선은 이전 스틸과 깔끔히 일치하지만, 계약이 명시한 '품에 안긴 채 은폐된 인물'이 통째로 사라져 샷 텍스트와 프레임 레이아웃의 핵심 요소를 충족하지 못한 실패작이다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "앉은 여성의 뒷모습이 화면 중앙에 있고, 몸통과 시선이 정면의 금빛 바다(석양 반사)를 향한다. 어깨선이 앞으로 숙여져 있어 품에 안은 대상 쪽으로 기울어져 있고, 카메라 축은 선착장 돌바닥의 직선 접근축을 따라 눈높이에서 뒤에서 바라본다. 안긴 인물의 머리는 앉은 여성의 오른쪽 어깨 너머 아래로 늘어져 있어 얼굴이 보이지 않는다.",
      "built_space": "선착장 돌바닥이 하단을 받치고, 좌측에 계류 로프·작은 배·난간 기둥, 우측에 통발 바구니 더미와 잡동사니, 화면 우상단으로 가로등 기둥 하나가 서 있다. 이전 스틸의 부두 끝·조명 기둥·통발 배치와 일치. 구조물 중복이나 잘못된 개수는 없음.",
      "entities": "앉은 인물은 40대 후반으로 보이는 한국 여성, 어두운 머리를 낮게 묶은 포니테일, 이전 스틸과 동일한 누비 카키 재킷·어두운 하의. 품에 안긴 인물은 젖은 듯 어두운 머리카락과 어두운 상의의 팔·손이 하단 우측으로 늘어져 부분만 보여, 신원 비노출 조건을 충족. 다만 안는 이의 왼쪽 종아리·맨발이 노출되어 이전 스틸의 복장(긴 바지)과 다소 어긋나며, 젖은 시신의 팔로도 읽힐 여지가 있는 애매한 다리 요소가 있다.",
      "hard_violations": [],
      "physics": "안긴 인물의 머리는 안는 이의 어깨/팔 위에 얹혀 중력대로 떨궈져 있고, 팔과 손은 돌바닥 쪽으로 완전히 늘어져 손등이 바닥에 닿아 있다. 근육 지지 없이 무릎과 팔로 받쳐지는 형태로, 지지점이 명확하다. 공중에 떠 있는 부분 없음. 다만 노출된 맨다리/발 하나가 누구 것인지 다소 모호한 해부학적 혼선이 있다."
     },
     {
      "label": "B",
      "direction": "동일한 뒷모습 앉은 자세, 몸과 시선은 금빛 바다를 향한다. 그러나 품에 안은 대상이 전혀 없고, 팔은 무릎 위에 얹혀 결가부좌에 가깝게 앉아 있다. 시선·자세가 향하는 목표는 바다뿐이며, 샷 텍스트가 요구한 '품에 안은 듯한 낯선 실루엣'의 대상이 화면에 존재하지 않는다.",
      "built_space": "선착장 돌바닥, 좌측 계류 로프·소형 배·난간, 우측 통발 더미, 우상단 기둥 — 이전 스틸과 재질·배치 모두 일치. 구조 오류 없음.",
      "entities": "앉은 인물은 40대 후반 한국 여성, 낮게 묶은 포니테일, 누비 카키 재킷·어두운 바지로 이전 스틸과 정확히 일치. 그러나 계약이 요구한 '안긴 채 은폐된 인물(강민숙의 시신)'이 아예 부재해 필수 엔티티 하나가 빠져 있다.",
      "hard_violations": [
       "샷 텍스트·프레임 레이아웃이 명시한 '품에 안긴 채 은폐된 인물(하단-중앙 미드그라운드)'이 전혀 렌더링되지 않음 — 필수 인물 누락"
      ],
      "physics": "여성은 돌바닥에 엉덩이와 다리로 완전히 앉아 있어 지지 문제 없음. 공중에 뜬 요소 없음. 다만 지지해야 할 시신 자체가 없어 물리 검증 대상이 성립하지 않는다."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "앉은 인물은 정확히 화면 중앙에서 등을 보이고 바다(금빛 석양)를 향해 있다. 시선 축은 프레임 안쪽 수평선·해 반사광 방향으로 정확히 향한다. 그러나 '품에 안은' 대상이 존재하지 않아 아무것도 향해 안겨 있지 않다.",
      "built_space": "선착장 돌바닥이 하단을 지지하고 좌측으로 계류 로프·작은 선체·말뚝, 우측으로 통발 더미와 전신주가 배치되어 이전 스틸의 재질과 시설이 유지된다. 구조물 개수·배치 이상 없음. 인물은 돌바닥 위에 실제로 앉아 있다.",
      "entities": "인물은 40대 후반 한국 여성으로 읽히는 어두운 머리 낮은 묶음, 이전 스틸과 동일한 베이지 누빔 재킷·어두운 바지. 그러나 계약이 요구한 '품에 안긴 익사한 여성(강민숙)'이 프레임 어디에도 없다 — 필수 요소 누락.",
      "hard_violations": [],
      "physics": "여성은 양반다리로 돌바닥에 완전히 접지되어 있고 무릎·엉덩이·손이 모두 지지된다. 공중에 뜬 신체는 없다. 다만 안긴 시신이 없으므로 '품에 안은 듯한 낯선 실루엣'이라는 형태 자체가 성립하지 않는다."
     },
     {
      "label": "B",
      "direction": "앉은 인물은 화면 중앙에서 등을 보이고 바다·해 반사광 쪽을 향한다. 무릎 위 오른쪽으로 안긴 인물의 머리가 넘어가 있고 그 상체·얼굴은 여자의 어깨 뒤로 가려져 신원이 드러나지 않는다. 안긴 몸의 방향은 바다 쪽 앉은 인물 쪽으로 정렬.",
      "built_space": "이전 스틸과 동일한 돌바닥, 좌측 계류 로프·말뚝·작은 배, 우측 통발 더미, 전신주가 A와 동일하게 유지되며 배경 변경 없음. 인물은 부두 끝 근처 돌바닥 위.",
      "entities": "안고 있는 인물: 이전 스틸의 베이지 누빔 재킷·어두운 바지 유지, 어두운 머리 낮은 묶음, 한국 여성. 안긴 인물: 젖은 듯 늘어진 어두운 긴 머리, 어두운 상의, 맨발·창백한 다리 — 익사한 강민숙으로 읽힌다. 다만 안는 인물의 오른쪽 맨다리/맨발이 재킷 아래로 노출되어 하의 연속성이 다소 흔들린다(경미).",
      "hard_violations": [],
      "physics": "안긴 몸은 앉은 인물의 무릎과 팔, 그리고 돌바닥에 걸쳐 완전히 지지된다. 머리는 근육 지지 없이 옆으로 축 늘어지고, 팔·손은 바닥으로 떨어져 늘어져 있으며 다리는 바닥에 닿는다 — 무중력 부양이나 들어올린 부위 없음, 중력 순응."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "이전 스틸을 거의 그대로 유지해 손실은 적지만, 샷 텍스트의 핵심인 '품에 안은 듯한 낯선 실루엣'과 무릎 위 시신(강민숙)이 완전히 빠져 계약의 중심 요소를 실현하지 못했다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "무릎·팔에 축 늘어진 시신이 얼굴을 가린 채 중력에 순응해 안겨 있어 샷 텍스트와 불변 포즈 계약을 충족하고, 장소·의상·광선도 이전 스틸과 일치한다; 안는 인물의 맨다리 노출이 유일한 경미한 손실."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "이전 스틸을 거의 그대로 유지해 손실은 적지만, 샷 텍스트의 핵심인 '품에 안은 듯한 낯선 실루엣'과 무릎 위 시신(강민숙)이 완전히 빠져 계약의 중심 요소를 실현하지 못했다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "무릎·팔에 축 늘어진 시신이 얼굴을 가린 채 중력에 순응해 안겨 있어 샷 텍스트와 불변 포즈 계약을 충족하고, 장소·의상·광선도 이전 스틸과 일치한다; 안는 인물의 맨다리 노출이 유일한 경미한 손실."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "앉은 인물은 정확히 화면 중앙에서 등을 보이고 바다(금빛 석양)를 향해 있다. 시선 축은 프레임 안쪽 수평선·해 반사광 방향으로 정확히 향한다. 그러나 '품에 안은' 대상이 존재하지 않아 아무것도 향해 안겨 있지 않다.",
      "built_space": "선착장 돌바닥이 하단을 지지하고 좌측으로 계류 로프·작은 선체·말뚝, 우측으로 통발 더미와 전신주가 배치되어 이전 스틸의 재질과 시설이 유지된다. 구조물 개수·배치 이상 없음. 인물은 돌바닥 위에 실제로 앉아 있다.",
      "entities": "인물은 40대 후반 한국 여성으로 읽히는 어두운 머리 낮은 묶음, 이전 스틸과 동일한 베이지 누빔 재킷·어두운 바지. 그러나 계약이 요구한 '품에 안긴 익사한 여성(강민숙)'이 프레임 어디에도 없다 — 필수 요소 누락.",
      "hard_violations": [],
      "physics": "여성은 양반다리로 돌바닥에 완전히 접지되어 있고 무릎·엉덩이·손이 모두 지지된다. 공중에 뜬 신체는 없다. 다만 안긴 시신이 없으므로 '품에 안은 듯한 낯선 실루엣'이라는 형태 자체가 성립하지 않는다."
     },
     {
      "label": "A",
      "direction": "앉은 인물은 화면 중앙에서 등을 보이고 바다·해 반사광 쪽을 향한다. 무릎 위 오른쪽으로 안긴 인물의 머리가 넘어가 있고 그 상체·얼굴은 여자의 어깨 뒤로 가려져 신원이 드러나지 않는다. 안긴 몸의 방향은 바다 쪽 앉은 인물 쪽으로 정렬.",
      "built_space": "이전 스틸과 동일한 돌바닥, 좌측 계류 로프·말뚝·작은 배, 우측 통발 더미, 전신주가 A와 동일하게 유지되며 배경 변경 없음. 인물은 부두 끝 근처 돌바닥 위.",
      "entities": "안고 있는 인물: 이전 스틸의 베이지 누빔 재킷·어두운 바지 유지, 어두운 머리 낮은 묶음, 한국 여성. 안긴 인물: 젖은 듯 늘어진 어두운 긴 머리, 어두운 상의, 맨발·창백한 다리 — 익사한 강민숙으로 읽힌다. 다만 안는 인물의 오른쪽 맨다리/맨발이 재킷 아래로 노출되어 하의 연속성이 다소 흔들린다(경미).",
      "hard_violations": [],
      "physics": "안긴 몸은 앉은 인물의 무릎과 팔, 그리고 돌바닥에 걸쳐 완전히 지지된다. 머리는 근육 지지 없이 옆으로 축 늘어지고, 팔·손은 바닥으로 떨어져 늘어져 있으며 다리는 바닥에 닿는다 — 무중력 부양이나 들어올린 부위 없음, 중력 순응."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 15,
     "B": 7
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S92sh3"
  },
  "locked_char_refs_excluded": [
   "강민숙(홍은미)(C04)"
  ]
 },
 "S101sh7": {
  "input_fingerprint": "12d40d63b92fb5d1",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 구창모가 미간을 잔뜩 찌푸린 채 심각하게 경고하는 얼굴 클로즈업.\n\nLOCATION (lock): The open deck beside the gangway of the fishing boat moored at the mainland pier. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: Resolve the dolly-in at 구창모's eye level on his side of the conversational axis, ending in a tight three-quarter facial close-up. His furrowed brow and clenched jaw occupy the center as he leans toward 황봉업(황 씨), who remains off-screen in the established reverse-angle direction.\n- FRAMING SCALE: close-up\n- KEY BACKGROUND ELEMENTS: 낚시 어선 (술집 여자들과 낚싯꾼들이 타고 있음) — Only a partial side and occupied upper area remain visible behind 구창모; used as A narrow contextual strip places the warning beside the crowded boat without distracting from 구창모's face; 선착장 (구창모와 황봉업이 대화하는 공간); used as Provides the immediate location of the conversation.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daytime illumination with firm facial contrast exposes the tension in 구창모's furrowed brow and lowered warning.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same crowded deck, bright sea, daylight, women, fishermen, and boatman nearby. Exclude the reveling passengers from the tight close-up; show the officer warning the boatman gravely.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (Korean man in his late 40s, short dark hair, lean sun-darkened face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 9,
   "C": 4
  },
  "selected": "B",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "시선과 몸 기울임이 오프스크린 황봉업에게 정확히 향하고 장소도 맞지만, 복장이 캐릭터 레퍼런스(비니·페이즐리 셔츠)와 어긋나고 배경이 다소 넓어 클로즈업 밀도가 B보다 약하다.  ★위반: [gpt] 오프스크린이어야 할 황봉업의 머리와 상체가 화면 오른쪽 전경에 등장한다. / [gpt] 구창모만 보여야 한다는 인물 제한과 승객 배제 지시를 어기고 배 위 여성들과 낚시꾼들의 신체가 다수 노출된다."
   },
   {
    "label": "B",
    "score": 9,
    "verdict_ko": "찌푸린 미간의 3/4 타이트 클로즈업, 눈높이 리버스 축, 오프스크린 상대를 향한 정확한 시선, 캐릭터 레퍼런스 의상(비니·페이즐리 셔츠)과 이전 스틸 장소가 모두 일치해 가장 충실하다.  ★위반: [gpt] 오프스크린이어야 할 황봉업의 머리와 어깨가 화면 오른쪽 전경에 등장한다. / [gpt] 구창모만 보여야 한다는 인물 제한과 승객 배제 지시를 어기고 배 위 여성들과 낚시꾼들의 신체가 다수 노출된다."
   },
   {
    "label": "C",
    "score": 4,
    "verdict_ko": "시선이 상대가 아닌 카메라 쪽으로 빠져 경고 대상이 착지하지 않고, 전경 버킷 더미와 넓은 배경으로 클로즈업 스케일도 어긋난다.  ★위반: [gpt] 오프스크린이어야 할 황봉업의 머리와 상체가 화면 왼쪽 전경에 크게 등장한다. / [gpt] 구창모만 보여야 한다는 인물 제한과 승객 배제 지시를 어기고 배 위 여성들과 낚시꾼들의 신체가 다수 노출된다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S101sh1_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:808030>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "황봉업은 화면 밖(off-screen)에 있어야 한다는 지시와 '구창모 외에 다른 인물을 추가하지 말라'는 규정을 위반하고 우측 전경에 인물의 뒷모습(머리와 어깨)이 크게 등장함.",
     "fix_en": "Remove the head and shoulder of the person in the right foreground completely, replacing that area with the continued blue sea and the concrete pier. Preserve Gu Chang-mo's exact pose, facial expression, clothing, and the background boat with all its passengers.",
     "needs_regeneration": false
    },
    {
     "issue_ko": "카메라 프레이밍이 지정된 '타이트한 얼굴 클로즈업(tight three-quarter facial close-up)'이 아니라 가슴까지 보이는 미디엄 클로즈업으로 설정되어, 프롬프트에서 프레임 밖으로 제외하라고 명시한 배경의 승객들이 크게 노출됨.",
     "fix_en": "No in-place fix is possible without relocating the camera to achieve the correct tight close-up.",
     "needs_regeneration": true
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the head and shoulder of the person in the right foreground completely, replacing that area with the continued blue sea and the concrete pier. Preserve Gu Chang-mo's exact pose, facial expression, clothing, and the background boat with all its passengers.\n- No in-place fix is possible without relocating the camera to achieve the correct tight close-up.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "구창모의 시선이 화면 오른쪽 전경에 있는 인물(황봉업으로 읽히는 남성의 뒤통수·어깨)을 정확히 향하고 있다. 미간을 찌푸리고 턱을 굳힌 채 상대 쪽으로 몸을 기울여 경고하는 순간이 성립한다. 배경의 여성들과 낚시꾼들은 각자 다른 방향(휴대폰, 바다)을 보고 있어 시선 충돌이 없다.",
      "built_space": "선착장과 계류된 어선의 조타실 흰 벽, 파란 차양, 확성기, 계선주와 로프, 난간이 레퍼런스와 일치한다. 조타실 문·창의 배열, 차양 프레임, 확성기 위치가 이전 스틸과 동일한 기하로 놓여 있고 구조물 중복은 없다. 배는 오른쪽 뒤로 이어지며 낚싯대와 뜰채가 일부만 보인다.",
      "entities": "구창모: 40대 후반 한국 남성, 짧은 머리 위 남색 비니, 회색 셔츠재킷 안 페이즐리 무늬 셔츠 — 캐릭터 레퍼런스 의상과 정확히 일치. 오른쪽 전경 인물은 뒷머리와 목·어깨만 보이는 성인 남성으로 이전 스틸의 선장(황봉업) 카키색 상의와 일치. 배경 여성 5명·낚시꾼들·아이스박스 등 이전 스틸 요소가 축소된 스케일로 유지됨.",
      "hard_violations": [],
      "physics": "구창모는 발이 프레임 밖이지만 상반신 무게가 앞으로 기운 자연스러운 대화 자세이며, 어깨·목 각도가 몸통과 연결되어 지지가 성립한다. 배경 인물들은 모두 갑판·부두 바닥에 서 있고 공중에 뜬 물체는 없다. 로프는 계선주에 감겨 있고 아이스박스는 부두면에 놓여 있다."
     },
     {
      "label": "B",
      "direction": "구창모의 시선은 화면 오른쪽 프레임 밖을 향하나, 그 방향에 대화 상대가 전혀 없다. 오른쪽 전경 인물이 제거되어 '황봉업 쪽으로 몸을 기울여 경고한다'는 관계가 화면상 대상 없이 허공을 보는 것으로 읽힌다. 배경 낚시꾼과 여성들의 방향은 A와 유사.",
      "built_space": "동일한 조타실·차양·확성기·계선주·로프가 보이나, 카메라가 뒤로 물러나 배 오른쪽이 잘려 나가고 바다 면적이 크게 늘었다. 부두 오른쪽 가장자리와 콘크리트 벽이 새로 드러나 프레임 오른쪽 절반이 빈 바다/부두로 채워졌다. 구조 중복은 없음.",
      "entities": "구창모: 동일한 비니·회색 셔츠재킷·페이즐리 셔츠로 레퍼런스와 일치. 다만 상대역(황봉업)의 신체 일부가 사라졌다. 배경 여성 5명, 낚시꾼 소수, 파란 아이스박스는 유지.",
      "hard_violations": [],
      "physics": "구창모는 상반신이 화면 중앙에 안정적으로 서 있고 지지에 문제 없음. 배경 인물들도 갑판에 접지. 공중에 뜬 물체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "'황봉업 쪽으로 기울여 경고하는' 대화축이 오른쪽 전경 어깨로 명확히 성립하고, 얼굴이 프레임을 채우는 3/4 타이트 클로즈업 요건과 이전 스틸의 장소·의상·배경 인물이 모두 유지된다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "동일 인물·의상·장소를 지키지만 카메라가 물러나 클로즈업 스케일이 약해지고 오른쪽 절반이 빈 바다로 바뀌었으며, 대화 상대(황봉업)의 존재감이 사라져 '상대를 향해 경고한다'는 지시 관계가 화면에서 소실된 것이 결정적 손실이다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "'황봉업 쪽으로 기울여 경고하는' 대화축이 오른쪽 전경 어깨로 명확히 성립하고, 얼굴이 프레임을 채우는 3/4 타이트 클로즈업 요건과 이전 스틸의 장소·의상·배경 인물이 모두 유지된다."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "동일 인물·의상·장소를 지키지만 카메라가 물러나 클로즈업 스케일이 약해지고 오른쪽 절반이 빈 바다로 바뀌었으며, 대화 상대(황봉업)의 존재감이 사라져 '상대를 향해 경고한다'는 지시 관계가 화면에서 소실된 것이 결정적 손실이다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "구창모의 시선이 화면 오른쪽 전경에 있는 인물(황봉업으로 읽히는 남성의 뒤통수·어깨)을 정확히 향하고 있다. 미간을 찌푸리고 턱을 굳힌 채 상대 쪽으로 몸을 기울여 경고하는 순간이 성립한다. 배경의 여성들과 낚시꾼들은 각자 다른 방향(휴대폰, 바다)을 보고 있어 시선 충돌이 없다.",
      "built_space": "선착장과 계류된 어선의 조타실 흰 벽, 파란 차양, 확성기, 계선주와 로프, 난간이 레퍼런스와 일치한다. 조타실 문·창의 배열, 차양 프레임, 확성기 위치가 이전 스틸과 동일한 기하로 놓여 있고 구조물 중복은 없다. 배는 오른쪽 뒤로 이어지며 낚싯대와 뜰채가 일부만 보인다.",
      "entities": "구창모: 40대 후반 한국 남성, 짧은 머리 위 남색 비니, 회색 셔츠재킷 안 페이즐리 무늬 셔츠 — 캐릭터 레퍼런스 의상과 정확히 일치. 오른쪽 전경 인물은 뒷머리와 목·어깨만 보이는 성인 남성으로 이전 스틸의 선장(황봉업) 카키색 상의와 일치. 배경 여성 5명·낚시꾼들·아이스박스 등 이전 스틸 요소가 축소된 스케일로 유지됨.",
      "hard_violations": [],
      "physics": "구창모는 발이 프레임 밖이지만 상반신 무게가 앞으로 기운 자연스러운 대화 자세이며, 어깨·목 각도가 몸통과 연결되어 지지가 성립한다. 배경 인물들은 모두 갑판·부두 바닥에 서 있고 공중에 뜬 물체는 없다. 로프는 계선주에 감겨 있고 아이스박스는 부두면에 놓여 있다."
     },
     {
      "label": "B",
      "direction": "구창모의 시선은 화면 오른쪽 프레임 밖을 향하나, 그 방향에 대화 상대가 전혀 없다. 오른쪽 전경 인물이 제거되어 '황봉업 쪽으로 몸을 기울여 경고한다'는 관계가 화면상 대상 없이 허공을 보는 것으로 읽힌다. 배경 낚시꾼과 여성들의 방향은 A와 유사.",
      "built_space": "동일한 조타실·차양·확성기·계선주·로프가 보이나, 카메라가 뒤로 물러나 배 오른쪽이 잘려 나가고 바다 면적이 크게 늘었다. 부두 오른쪽 가장자리와 콘크리트 벽이 새로 드러나 프레임 오른쪽 절반이 빈 바다/부두로 채워졌다. 구조 중복은 없음.",
      "entities": "구창모: 동일한 비니·회색 셔츠재킷·페이즐리 셔츠로 레퍼런스와 일치. 다만 상대역(황봉업)의 신체 일부가 사라졌다. 배경 여성 5명, 낚시꾼 소수, 파란 아이스박스는 유지.",
      "hard_violations": [],
      "physics": "구창모는 상반신이 화면 중앙에 안정적으로 서 있고 지지에 문제 없음. 배경 인물들도 갑판에 접지. 공중에 뜬 물체 없음."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "구창모는 화면 오른쪽(카메라 우측 off-screen)을 향해 시선을 낮게 던지며 경고하는 표정. 눈길이 프레임 밖 대화 상대를 향해 정확히 나가며, 리버스 앵글 방향(황봉업이 화면 밖 우측)과 일치. 몸통도 약간 우측으로 기울어 대화축 위에 놓임.",
      "built_space": "배경은 어선의 조타실 흰 벽, 파란 차양, 확성기, 난간, 계선주와 로프, 그리고 선착장 콘크리트 바닥 — 이전 스틸의 고정물과 재질이 일치. 어선 측면과 낚시꾼이 있는 상부만 좁은 띠로 남아 있어 계약이 요구한 'narrow contextual strip'에 부합. 구조물 중복 없음.",
      "entities": "구창모: 40대 후반 한국인 남성, 그을린 마른 얼굴, 네이비 비니, 회색 셔츠재킷 위 페이즐리 셔츠 — 캐릭터 레퍼런스와 일치. 배경의 술집 여자 5인, 낚시꾼들, 파란 아이스박스 등 이전 스틸 요소가 축소된 실제 크기로 유지. 새로 발명된 인물 없음.",
      "hard_violations": [],
      "physics": "구창모는 선착장 위에 두 발로 선 채 상반신만 프레임 안. 어깨선과 목의 긴장이 앞으로 기울어진 자세를 뒷받침. 공중에 뜬 물체나 지지 없는 신체 없음. 배경 인물들도 갑판/부두에 접지."
     },
     {
      "label": "B",
      "direction": "구창모의 시선은 화면 오른쪽 상대의 얼굴 쪽으로 향함. 다만 프레임 우측에 상대 남자의 뒤통수와 어깨가 크게 들어와 오버더숄더 구도가 되어, 계약이 명시한 '상대는 off-screen' 조건과 어긋남.",
      "built_space": "동일한 조타실, 차양, 확성기, 계선주 로프, 파란 아이스박스 — 장소는 일치. 그러나 우측 배경(바다·어선 우현)이 인물 뒤통수에 가려짐. 구조 중복은 없음.",
      "entities": "구창모 동일 인물·의상 일치. 추가로 등장한 우측 인물은 짧은 검은 머리와 카키색 상의로 황봉업으로 읽히나, 계약은 'never add a person the shot text does not show'이며 상대는 off-screen이어야 함 — 계약이 요구하지 않은 인물 추가.",
      "hard_violations": [
       "샷 텍스트·카메라 지시가 황봉업을 off-screen으로 명시했음에도 상대 인물의 머리·어깨를 프레임에 크게 넣어 오버더숄더로 만든 인물 추가(PEOPLE 조항: 구창모 외 인물 금지)"
      ],
      "physics": "구창모는 부두 위에 서 있고 접지 문제 없음. 우측 인물도 프레임 하단으로 이어지는 몸통이 있어 부유 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "황봉업을 화면 밖에 둔 채 구창모의 3/4 얼굴 클로즈업을 정확히 실현했고, 배경 어선·여성들·낚시꾼이 좁은 맥락 띠로만 남아 계약의 프레이밍·인물 제한을 모두 충족."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "상대 인물의 뒤통수·어깨를 크게 넣어 오버더숄더로 바꾼 것은 'remains off-screen' 및 '구창모 외 인물 금지' 조항을 정면으로 위반하고, 그 대가로 우측 배경(바다·어선)까지 가려 얻은 것보다 잃은 것이 큼."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "황봉업을 화면 밖에 둔 채 구창모의 3/4 얼굴 클로즈업을 정확히 실현했고, 배경 어선·여성들·낚시꾼이 좁은 맥락 띠로만 남아 계약의 프레이밍·인물 제한을 모두 충족."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "상대 인물의 뒤통수·어깨를 크게 넣어 오버더숄더로 바꾼 것은 'remains off-screen' 및 '구창모 외 인물 금지' 조항을 정면으로 위반하고, 그 대가로 우측 배경(바다·어선)까지 가려 얻은 것보다 잃은 것이 큼."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "구창모는 화면 오른쪽(카메라 우측 off-screen)을 향해 시선을 낮게 던지며 경고하는 표정. 눈길이 프레임 밖 대화 상대를 향해 정확히 나가며, 리버스 앵글 방향(황봉업이 화면 밖 우측)과 일치. 몸통도 약간 우측으로 기울어 대화축 위에 놓임.",
      "built_space": "배경은 어선의 조타실 흰 벽, 파란 차양, 확성기, 난간, 계선주와 로프, 그리고 선착장 콘크리트 바닥 — 이전 스틸의 고정물과 재질이 일치. 어선 측면과 낚시꾼이 있는 상부만 좁은 띠로 남아 있어 계약이 요구한 'narrow contextual strip'에 부합. 구조물 중복 없음.",
      "entities": "구창모: 40대 후반 한국인 남성, 그을린 마른 얼굴, 네이비 비니, 회색 셔츠재킷 위 페이즐리 셔츠 — 캐릭터 레퍼런스와 일치. 배경의 술집 여자 5인, 낚시꾼들, 파란 아이스박스 등 이전 스틸 요소가 축소된 실제 크기로 유지. 새로 발명된 인물 없음.",
      "hard_violations": [],
      "physics": "구창모는 선착장 위에 두 발로 선 채 상반신만 프레임 안. 어깨선과 목의 긴장이 앞으로 기울어진 자세를 뒷받침. 공중에 뜬 물체나 지지 없는 신체 없음. 배경 인물들도 갑판/부두에 접지."
     },
     {
      "label": "A",
      "direction": "구창모의 시선은 화면 오른쪽 상대의 얼굴 쪽으로 향함. 다만 프레임 우측에 상대 남자의 뒤통수와 어깨가 크게 들어와 오버더숄더 구도가 되어, 계약이 명시한 '상대는 off-screen' 조건과 어긋남.",
      "built_space": "동일한 조타실, 차양, 확성기, 계선주 로프, 파란 아이스박스 — 장소는 일치. 그러나 우측 배경(바다·어선 우현)이 인물 뒤통수에 가려짐. 구조 중복은 없음.",
      "entities": "구창모 동일 인물·의상 일치. 추가로 등장한 우측 인물은 짧은 검은 머리와 카키색 상의로 황봉업으로 읽히나, 계약은 'never add a person the shot text does not show'이며 상대는 off-screen이어야 함 — 계약이 요구하지 않은 인물 추가.",
      "hard_violations": [
       "샷 텍스트·카메라 지시가 황봉업을 off-screen으로 명시했음에도 상대 인물의 머리·어깨를 프레임에 크게 넣어 오버더숄더로 만든 인물 추가(PEOPLE 조항: 구창모 외 인물 금지)"
      ],
      "physics": "구창모는 부두 위에 서 있고 접지 문제 없음. 우측 인물도 프레임 하단으로 이어지는 몸통이 있어 부유 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 13
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": false,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S101sh1"
  }
 },
 "S102sh1::bgfirst_bg": {
  "input_fingerprint": "3c1d37b75a144590",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 동굴 안에서 정인우가 핏대를 세운 채 벽에 연결된 굵은 쇠사슬을 양손으로 팽팽하게 잡아당기는 얼굴 클로즈업.\n\nLOCATION (lock): The rock wall and chained-prisoner position inside the dark cave holding chamber.\n\nTIME OF DAY (lock): day transitioning through dusk to evening.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At face height within arm’s reach, the camera faces 정인우 beside the wall connection point and holds the opening position of the planned dolly-out. His strained face occupies the center and upper half, while both pulling hands and the taut chain enter across the lower frame, keeping the anchor visible near one edge.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 정인우 in the middle-center of the frame, foreground; wall connection point in the middle-right of the frame, midground.\n- KEY BACKGROUND ELEMENTS: wall connection point (Connected to the taut chain) — Its projecting side and the attached chain face the camera at the frame edge; used as Keeps the chain’s fixed endpoint visible beside the straining face; thick iron chain (Pulled taut) — Its length angles from the lower frame toward the wall connection point; used as Forms a hard diagonal from the hands toward the wall.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient light preserves the darkness of the daytime cave while sharply modeling the strain across his face and hands.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 동굴 안에서 정인우가 핏대를 세운 채 벽에 연결된 굵은 쇠사슬을 양손으로 팽팽하게 잡아당기는 얼굴 클로즈업.\n\nLOCATION (lock): The rock wall and chained-prisoner position inside the dark cave holding chamber.\n\nTIME OF DAY (lock): day transitioning through dusk to evening.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At face height within arm’s reach, the camera faces 정인우 beside the wall connection point and holds the opening position of the planned dolly-out. His strained face occupies the center and upper half, while both pulling hands and the taut chain enter across the lower frame, keeping the anchor visible near one edge.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 정인우 in the middle-center of the frame, foreground; wall connection point in the middle-right of the frame, midground.\n- KEY BACKGROUND ELEMENTS: wall connection point (Connected to the taut chain) — Its projecting side and the attached chain face the camera at the frame edge; used as Keeps the chain’s fixed endpoint visible beside the straining face; thick iron chain (Pulled taut) — Its length angles from the lower frame toward the wall connection point; used as Forms a hard diagonal from the hands toward the wall.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient light preserves the darkness of the daytime cave while sharply modeling the strain across his face and hands.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S102sh1__bgfirst_bg.png",
  "asset_id": "3ff55a5d-1ae5-489b-93c4-01f57e4578e9",
  "input_asset_ids": [
   "6a9be857-71c9-4e63-a0e4-a5189b2b4356",
   "92d7c07c-a1e6-4db3-8f01-1a1b44de8540"
  ]
 },
 "S102sh1": {
  "input_fingerprint": "9dfc785984a4b8dc",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day transitioning through dusk to evening.\n\nSHOT TEXT (authoritative, Korean): 어두운 동굴 안에서 정인우가 핏대를 세운 채 벽에 연결된 굵은 쇠사슬을 양손으로 팽팽하게 잡아당기는 얼굴 클로즈업.\n\nLOCATION (lock): The rock wall and chained-prisoner position inside the dark cave holding chamber. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At face height within arm’s reach, the camera faces 정인우 beside the wall connection point and holds the opening position of the planned dolly-out. His strained face occupies the center and upper half, while both pulling hands and the taut chain enter across the lower frame, keeping the anchor visible near one edge.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 정인우 in the middle-center of the frame, foreground; wall connection point in the middle-right of the frame, midground.\n- KEY BACKGROUND ELEMENTS: wall connection point (Connected to the taut chain) — Its projecting side and the attached chain face the camera at the frame edge; used as Keeps the chain’s fixed endpoint visible beside the straining face; thick iron chain (Pulled taut) — Its length angles from the lower frame toward the wall connection point; used as Forms a hard diagonal from the hands toward the wall.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient light preserves the darkness of the daytime cave while sharply modeling the strain across his face and hands.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo and Jihu remain chained to the cave walls; his injured arm remains casted, and her necklace remains at the ritual altar.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 정인우 right now, so 정인우's hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 정인우: its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day transitioning through dusk to evening.\n\nSHOT TEXT (authoritative, Korean): 어두운 동굴 안에서 정인우가 핏대를 세운 채 벽에 연결된 굵은 쇠사슬을 양손으로 팽팽하게 잡아당기는 얼굴 클로즈업.\n\nLOCATION (lock): The rock wall and chained-prisoner position inside the dark cave holding chamber. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At face height within arm’s reach, the camera faces 정인우 beside the wall connection point and holds the opening position of the planned dolly-out. His strained face occupies the center and upper half, while both pulling hands and the taut chain enter across the lower frame, keeping the anchor visible near one edge.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 정인우 in the middle-center of the frame, foreground; wall connection point in the middle-right of the frame, midground.\n- KEY BACKGROUND ELEMENTS: wall connection point (Connected to the taut chain) — Its projecting side and the attached chain face the camera at the frame edge; used as Keeps the chain’s fixed endpoint visible beside the straining face; thick iron chain (Pulled taut) — Its length angles from the lower frame toward the wall connection point; used as Forms a hard diagonal from the hands toward the wall.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient light preserves the darkness of the daytime cave while sharply modeling the strain across his face and hands.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo and Jihu remain chained to the cave walls; his injured arm remains casted, and her necklace remains at the ritual altar.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 정인우 right now, so 정인우's hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 정인우: its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day transitioning through dusk to evening.\n\nSHOT TEXT (authoritative, Korean): 어두운 동굴 안에서 정인우가 핏대를 세운 채 벽에 연결된 굵은 쇠사슬을 양손으로 팽팽하게 잡아당기는 얼굴 클로즈업.\n\nLOCATION (lock): The rock wall and chained-prisoner position inside the dark cave holding chamber. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At face height within arm’s reach, the camera faces 정인우 beside the wall connection point and holds the opening position of the planned dolly-out. His strained face occupies the center and upper half, while both pulling hands and the taut chain enter across the lower frame, keeping the anchor visible near one edge.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 정인우 in the middle-center of the frame, foreground; wall connection point in the middle-right of the frame, midground.\n- KEY BACKGROUND ELEMENTS: wall connection point (Connected to the taut chain) — Its projecting side and the attached chain face the camera at the frame edge; used as Keeps the chain’s fixed endpoint visible beside the straining face; thick iron chain (Pulled taut) — Its length angles from the lower frame toward the wall connection point; used as Forms a hard diagonal from the hands toward the wall.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key ambient light preserves the darkness of the daytime cave while sharply modeling the strain across his face and hands.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo and Jihu remain chained to the cave walls; his injured arm remains casted, and her necklace remains at the ritual altar.\n\nTHE HAND THAT IS DOING THIS: the object at the centre of this shot is being held, operated, read, aimed or handed over by 정인우 right now, so 정인우's hand — and as much of the wrist and forearm as the framing reaches — is in the frame, gripping or resting on that object exactly the way its real use requires. Match that hand to 정인우: its size, build, skin, age, grooming, sleeve and anything worn on it belong to that person and to no one else. Nothing else of them needs to be in shot; the framing stays on the object.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S102sh1__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S102sh1.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:787232>"
    },
    {
     "label": "PROP REFERENCE — 굵은 쇠사슬: the exact object appearing in this shot; match its look, material and wear exactly.",
     "path": "<bytes:963202>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L37B06.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:787232>"
    },
    {
     "label": "PROP REFERENCE — 굵은 쇠사슬: the exact object appearing in this shot; match its look, material and wear exactly.",
     "path": "<bytes:963202>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "ranking": [
     "B",
     "A"
    ],
    "winner": "B",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "정인우의 시선은 카메라 쪽(약간 아래·정면)을 향하고, 양손은 화면 왼쪽 아래에서 카메라 쪽으로 뻗어 쇠사슬을 쥐고 있다. 사슬은 손에서 오른쪽 상단의 벽 고정 링(프레임 우측 중단)으로 대각선을 이루며 이어져, 텍스트가 요구한 '손→벽 연결점' 대각선은 성립한다. 다만 얼굴이 화면 중앙 상단이 아니라 좌상단으로 밀려 있고, 시선이 렌즈를 정면으로 응시해 '당기는 힘의 방향'과 다소 어긋난다.",
      "built_space": "동굴 암벽과 안쪽으로 이어지는 갱도, 천장의 작은 조명이 로케이션 레퍼런스와 일치한다. 벽 고정 링은 우측 암벽에 박혀 있고 사슬이 거기서 나온다 — 개수(고리 1개, 사슬 1줄)도 맞다. 다만 인물이 벽에서 상당히 떨어져 서 있어 '벽에 연결된 죄수 위치'라기보다 통로 한가운데 서 있는 느낌이며, 프레이밍이 클로즈업이 아니라 미디엄 클로즈업~바스트에 가깝다.",
      "entities": "한국인 남성, 검은 머리이나 얼굴 연령대가 30대 후반~40대로 읽혀 '20대 중반의 앳되고 잘생긴 얼굴'이라는 캐릭터 레퍼런스와 상당히 어긋난다. 의상은 어두운 셔츠. 쇠사슬은 굵기가 레퍼런스보다 가늘고 링크가 얇아 '굵은 쇠사슬'로는 다소 약하다. 팔에 깁스는 보이지 않는다.",
      "hard_violations": [],
      "physics": "두 손이 모두 사슬을 실제로 감아쥐고 있고 팔뚝 힘줄·전완 근육이 긴장되어 있다. 사슬은 손과 벽 고정점 사이에서 팽팽하며 처짐이 없다. 몸은 지면에 서 있고 지지가 명확하다. 다만 팔이 카메라 쪽으로 뻗어 있어 '벽 쪽으로 당기는' 반작용 방향은 성립하나, 손이 광각 왜곡으로 과장되게 커 보인다."
     },
     {
      "label": "B",
      "direction": "인물의 시선과 얼굴은 화면 오른쪽 벽 고정점(사슬이 박힌 지점) 쪽을 노려보고 있고, 양손은 화면 하단을 가로질러 사슬을 쥔 채 몸 쪽으로 당긴다. 사슬은 왼쪽 아래에서 오른쪽 위 벽 고정 브래킷으로 뚜렷한 대각선을 형성해 '손 → 벽 연결점' 대각선 요구를 정확히 충족한다. 시선 타깃(사슬/고정점)이 행동과 일치한다.",
      "built_space": "암벽 클로즈업과 뒤편 어두운 갱도 입구가 로케이션과 부합한다. 벽 연결점은 프레임 우측 중단에 볼트로 고정된 아이플레이트 1개, 거기서 사슬 1줄이 나온다 — 개수·배치 모두 맞다. 인물이 벽 바로 옆에 밀착해 있어 '벽에 사슬로 묶인 죄수 위치'가 성립하고, 스케일도 과장 없이 실제 크기로 읽힌다. 프레이밍은 요구된 얼굴 클로즈업(얼굴이 중앙·상반 차지, 손과 사슬이 하단 가로지름)에 정확히 부합한다.",
      "entities": "한국인 남성, 검은 머리, 20대 후반으로 읽히는 젊은 얼굴로 캐릭터 레퍼런스에 A보다 가깝다. 의상은 흙 묻은 회백색 셔츠(감금 상태에 부합). 사슬은 굵은 단조 링크로 레퍼런스의 두께·마모감과 유사하다. 깁스는 소매에 가려 보이지 않으나 프레이밍상 감점 대상 아님.",
      "hard_violations": [],
      "physics": "오른손은 벽 고정점 근처 사슬을, 왼손은 그 아래 사슬을 실제로 움켜쥐고 있으며 손등 힘줄과 관절이 하얗게 힘이 들어가 있다. 사슬은 두 손 사이와 고정점까지 처짐 없이 팽팽하다. 몸은 벽에 기대며 뒤로 체중을 실어 당기는 자세로, 지지와 반작용이 명확하다. 공중에 뜬 요소 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "사슬–벽 고정점 대각선은 살렸으나 프레이밍이 클로즈업이 아닌 바스트급이고 얼굴이 중앙 상단이 아닌 좌상단으로 밀렸으며, 인물 연령대가 20대 중반 캐릭터 레퍼런스와 크게 어긋나고 사슬 굵기도 레퍼런스보다 가늘다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "얼굴이 프레임 중앙·상반을 차지하고 양손과 팽팽한 사슬이 하단을 가로지르며 우측 벽 연결점으로 대각선을 이루는 지정 구도를 정확히 구현했고, 저조도 동굴 톤·굵은 사슬 재질·젊은 인물 인상 모두 레퍼런스에 부합한다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "사슬–벽 고정점 대각선은 살렸으나 프레이밍이 클로즈업이 아닌 바스트급이고 얼굴이 중앙 상단이 아닌 좌상단으로 밀렸으며, 인물 연령대가 20대 중반 캐릭터 레퍼런스와 크게 어긋나고 사슬 굵기도 레퍼런스보다 가늘다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "얼굴이 프레임 중앙·상반을 차지하고 양손과 팽팽한 사슬이 하단을 가로지르며 우측 벽 연결점으로 대각선을 이루는 지정 구도를 정확히 구현했고, 저조도 동굴 톤·굵은 사슬 재질·젊은 인물 인상 모두 레퍼런스에 부합한다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "정인우의 시선은 카메라 쪽(약간 아래·정면)을 향하고, 양손은 화면 왼쪽 아래에서 카메라 쪽으로 뻗어 쇠사슬을 쥐고 있다. 사슬은 손에서 오른쪽 상단의 벽 고정 링(프레임 우측 중단)으로 대각선을 이루며 이어져, 텍스트가 요구한 '손→벽 연결점' 대각선은 성립한다. 다만 얼굴이 화면 중앙 상단이 아니라 좌상단으로 밀려 있고, 시선이 렌즈를 정면으로 응시해 '당기는 힘의 방향'과 다소 어긋난다.",
      "built_space": "동굴 암벽과 안쪽으로 이어지는 갱도, 천장의 작은 조명이 로케이션 레퍼런스와 일치한다. 벽 고정 링은 우측 암벽에 박혀 있고 사슬이 거기서 나온다 — 개수(고리 1개, 사슬 1줄)도 맞다. 다만 인물이 벽에서 상당히 떨어져 서 있어 '벽에 연결된 죄수 위치'라기보다 통로 한가운데 서 있는 느낌이며, 프레이밍이 클로즈업이 아니라 미디엄 클로즈업~바스트에 가깝다.",
      "entities": "한국인 남성, 검은 머리이나 얼굴 연령대가 30대 후반~40대로 읽혀 '20대 중반의 앳되고 잘생긴 얼굴'이라는 캐릭터 레퍼런스와 상당히 어긋난다. 의상은 어두운 셔츠. 쇠사슬은 굵기가 레퍼런스보다 가늘고 링크가 얇아 '굵은 쇠사슬'로는 다소 약하다. 팔에 깁스는 보이지 않는다.",
      "hard_violations": [],
      "physics": "두 손이 모두 사슬을 실제로 감아쥐고 있고 팔뚝 힘줄·전완 근육이 긴장되어 있다. 사슬은 손과 벽 고정점 사이에서 팽팽하며 처짐이 없다. 몸은 지면에 서 있고 지지가 명확하다. 다만 팔이 카메라 쪽으로 뻗어 있어 '벽 쪽으로 당기는' 반작용 방향은 성립하나, 손이 광각 왜곡으로 과장되게 커 보인다."
     },
     {
      "label": "B",
      "direction": "인물의 시선과 얼굴은 화면 오른쪽 벽 고정점(사슬이 박힌 지점) 쪽을 노려보고 있고, 양손은 화면 하단을 가로질러 사슬을 쥔 채 몸 쪽으로 당긴다. 사슬은 왼쪽 아래에서 오른쪽 위 벽 고정 브래킷으로 뚜렷한 대각선을 형성해 '손 → 벽 연결점' 대각선 요구를 정확히 충족한다. 시선 타깃(사슬/고정점)이 행동과 일치한다.",
      "built_space": "암벽 클로즈업과 뒤편 어두운 갱도 입구가 로케이션과 부합한다. 벽 연결점은 프레임 우측 중단에 볼트로 고정된 아이플레이트 1개, 거기서 사슬 1줄이 나온다 — 개수·배치 모두 맞다. 인물이 벽 바로 옆에 밀착해 있어 '벽에 사슬로 묶인 죄수 위치'가 성립하고, 스케일도 과장 없이 실제 크기로 읽힌다. 프레이밍은 요구된 얼굴 클로즈업(얼굴이 중앙·상반 차지, 손과 사슬이 하단 가로지름)에 정확히 부합한다.",
      "entities": "한국인 남성, 검은 머리, 20대 후반으로 읽히는 젊은 얼굴로 캐릭터 레퍼런스에 A보다 가깝다. 의상은 흙 묻은 회백색 셔츠(감금 상태에 부합). 사슬은 굵은 단조 링크로 레퍼런스의 두께·마모감과 유사하다. 깁스는 소매에 가려 보이지 않으나 프레이밍상 감점 대상 아님.",
      "hard_violations": [],
      "physics": "오른손은 벽 고정점 근처 사슬을, 왼손은 그 아래 사슬을 실제로 움켜쥐고 있으며 손등 힘줄과 관절이 하얗게 힘이 들어가 있다. 사슬은 두 손 사이와 고정점까지 처짐 없이 팽팽하다. 몸은 벽에 기대며 뒤로 체중을 실어 당기는 자세로, 지지와 반작용이 명확하다. 공중에 뜬 요소 없음."
     }
    ]
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "얼굴이 프레임 중앙·상반부를 차지하는 진짜 클로즈업이며, 두 손이 하단으로 들어와 쇠사슬이 우측 벽 고정점(브래킷+링)까지 팽팽한 대각선을 이루어 카메라·프레임 지시를 가장 정확히 충족한다. 사슬 링크가 레퍼런스보다 다소 가는 점은 사소한 차이."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "동굴 로케이션은 잘 살렸으나 미디엄 클로즈업으로 물러나 얼굴 비중이 작고, 오른손은 사슬을 실제로 감싸 쥐지 않은 채 뜬 주먹이며 왼손 그립도 불명확해 '양손으로 팽팽히 당김'과 클로즈업 스케일 모두 약하다. 배경 갱도 조명이 전경화된 점도 감점."
     }
    ],
    "readings": [
     {
      "label": "A",
      "direction": "인물의 시선은 카메라 축을 살짝 벗어난 정면(당기는 힘의 방향)으로 향하고, 사슬은 화면 좌하단에서 우측 벽의 금속 브래킷·링으로 곧게 이어져 명확한 대각선을 만든다. 사슬의 종착점이 우측 프레임 가장자리 중단에 정확히 보인다.",
      "built_space": "동굴 암벽이 우측 근경, 좌측 뒤로 어두운 갱도 공간이 열려 있어 레퍼런스의 암반 질감·저조도와 부합. 벽 고정점은 볼트 플레이트+아이링 1개로 중복 없이 올바른 수량.",
      "entities": "20대 중후반 한국 남성, 검은 단발머리, 흙과 땀이 묻은 회색 리넨 셔츠. 캐릭터 레퍼런스의 이목구비와 대체로 일치(다소 나이 들어 보임). 굵은 쇠사슬은 단조 철제 질감이나 레퍼런스 대비 링크가 약간 가늘다.",
      "hard_violations": [],
      "physics": "양손이 실제로 사슬 링크를 감아쥐고 있고 손등 힘줄·전완 근육이 긴장, 상체가 사슬 반대 방향으로 기울어 인장력이 성립. 사슬은 벽 앵커에 걸려 팽팽하게 지지됨. 부유하는 물체 없음."
     },
     {
      "label": "B",
      "direction": "시선은 렌즈를 거의 정면으로 응시하며, 사슬은 우측 벽의 아이볼트 링에서 좌하단 손 쪽으로 대각선을 그린다. 고정점은 우측 중단에 보이나 프레임 안쪽으로 크게 들어와 있고, 배경 갱도 조명이 중앙 뒤에 도드라진다.",
      "built_space": "암반 벽과 뒤쪽 갱도·전구가 레퍼런스 로케이션과 유사. 벽 고정 아이볼트 1개로 수량은 정상이나, 원경 통로가 실제보다 크게 밝게 강조되어 전경화 경향.",
      "entities": "한국 남성이나 30~40대로 보여 '20대 중반'과 거리가 있고, 검은 셔츠·목과 얼굴의 긁힌 상처. 쇠사슬 링크 굵기는 레퍼런스에 가깝다.",
      "hard_violations": [],
      "physics": "왼손은 사슬 하단 링크를 어느 정도 쥐었으나, 오른손은 사슬 위·앞쪽에서 링크를 감싸지 못한 채 빈 주먹 상태로 떠 있어 인장 전달이 불분명. 몸통은 정면으로 서 있어 당기는 반작용(체중 뒤로 실림)이 거의 보이지 않는다. 그 외 부유 물체는 없음."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "얼굴이 프레임 중앙·상반부를 차지하는 진짜 클로즈업이며, 두 손이 하단으로 들어와 쇠사슬이 우측 벽 고정점(브래킷+링)까지 팽팽한 대각선을 이루어 카메라·프레임 지시를 가장 정확히 충족한다. 사슬 링크가 레퍼런스보다 다소 가는 점은 사소한 차이."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "동굴 로케이션은 잘 살렸으나 미디엄 클로즈업으로 물러나 얼굴 비중이 작고, 오른손은 사슬을 실제로 감싸 쥐지 않은 채 뜬 주먹이며 왼손 그립도 불명확해 '양손으로 팽팽히 당김'과 클로즈업 스케일 모두 약하다. 배경 갱도 조명이 전경화된 점도 감점."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "인물의 시선은 카메라 축을 살짝 벗어난 정면(당기는 힘의 방향)으로 향하고, 사슬은 화면 좌하단에서 우측 벽의 금속 브래킷·링으로 곧게 이어져 명확한 대각선을 만든다. 사슬의 종착점이 우측 프레임 가장자리 중단에 정확히 보인다.",
      "built_space": "동굴 암벽이 우측 근경, 좌측 뒤로 어두운 갱도 공간이 열려 있어 레퍼런스의 암반 질감·저조도와 부합. 벽 고정점은 볼트 플레이트+아이링 1개로 중복 없이 올바른 수량.",
      "entities": "20대 중후반 한국 남성, 검은 단발머리, 흙과 땀이 묻은 회색 리넨 셔츠. 캐릭터 레퍼런스의 이목구비와 대체로 일치(다소 나이 들어 보임). 굵은 쇠사슬은 단조 철제 질감이나 레퍼런스 대비 링크가 약간 가늘다.",
      "hard_violations": [],
      "physics": "양손이 실제로 사슬 링크를 감아쥐고 있고 손등 힘줄·전완 근육이 긴장, 상체가 사슬 반대 방향으로 기울어 인장력이 성립. 사슬은 벽 앵커에 걸려 팽팽하게 지지됨. 부유하는 물체 없음."
     },
     {
      "label": "A",
      "direction": "시선은 렌즈를 거의 정면으로 응시하며, 사슬은 우측 벽의 아이볼트 링에서 좌하단 손 쪽으로 대각선을 그린다. 고정점은 우측 중단에 보이나 프레임 안쪽으로 크게 들어와 있고, 배경 갱도 조명이 중앙 뒤에 도드라진다.",
      "built_space": "암반 벽과 뒤쪽 갱도·전구가 레퍼런스 로케이션과 유사. 벽 고정 아이볼트 1개로 수량은 정상이나, 원경 통로가 실제보다 크게 밝게 강조되어 전경화 경향.",
      "entities": "한국 남성이나 30~40대로 보여 '20대 중반'과 거리가 있고, 검은 셔츠·목과 얼굴의 긁힌 상처. 쇠사슬 링크 굵기는 레퍼런스에 가깝다.",
      "hard_violations": [],
      "physics": "왼손은 사슬 하단 링크를 어느 정도 쥐었으나, 오른손은 사슬 위·앞쪽에서 링크를 감싸지 못한 채 빈 주먹 상태로 떠 있어 인장 전달이 불분명. 몸통은 정면으로 서 있어 당기는 반작용(체중 뒤로 실림)이 거의 보이지 않는다. 그 외 부유 물체는 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 10,
     "B": 17
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "readings": [
   {
    "label": "A",
    "direction": "정인우의 시선은 카메라 쪽(약간 아래·정면)을 향하고, 양손은 화면 왼쪽 아래에서 카메라 쪽으로 뻗어 쇠사슬을 쥐고 있다. 사슬은 손에서 오른쪽 상단의 벽 고정 링(프레임 우측 중단)으로 대각선을 이루며 이어져, 텍스트가 요구한 '손→벽 연결점' 대각선은 성립한다. 다만 얼굴이 화면 중앙 상단이 아니라 좌상단으로 밀려 있고, 시선이 렌즈를 정면으로 응시해 '당기는 힘의 방향'과 다소 어긋난다.",
    "built_space": "동굴 암벽과 안쪽으로 이어지는 갱도, 천장의 작은 조명이 로케이션 레퍼런스와 일치한다. 벽 고정 링은 우측 암벽에 박혀 있고 사슬이 거기서 나온다 — 개수(고리 1개, 사슬 1줄)도 맞다. 다만 인물이 벽에서 상당히 떨어져 서 있어 '벽에 연결된 죄수 위치'라기보다 통로 한가운데 서 있는 느낌이며, 프레이밍이 클로즈업이 아니라 미디엄 클로즈업~바스트에 가깝다.",
    "entities": "한국인 남성, 검은 머리이나 얼굴 연령대가 30대 후반~40대로 읽혀 '20대 중반의 앳되고 잘생긴 얼굴'이라는 캐릭터 레퍼런스와 상당히 어긋난다. 의상은 어두운 셔츠. 쇠사슬은 굵기가 레퍼런스보다 가늘고 링크가 얇아 '굵은 쇠사슬'로는 다소 약하다. 팔에 깁스는 보이지 않는다.",
    "hard_violations": [],
    "physics": "두 손이 모두 사슬을 실제로 감아쥐고 있고 팔뚝 힘줄·전완 근육이 긴장되어 있다. 사슬은 손과 벽 고정점 사이에서 팽팽하며 처짐이 없다. 몸은 지면에 서 있고 지지가 명확하다. 다만 팔이 카메라 쪽으로 뻗어 있어 '벽 쪽으로 당기는' 반작용 방향은 성립하나, 손이 광각 왜곡으로 과장되게 커 보인다."
   },
   {
    "label": "B",
    "direction": "인물의 시선과 얼굴은 화면 오른쪽 벽 고정점(사슬이 박힌 지점) 쪽을 노려보고 있고, 양손은 화면 하단을 가로질러 사슬을 쥔 채 몸 쪽으로 당긴다. 사슬은 왼쪽 아래에서 오른쪽 위 벽 고정 브래킷으로 뚜렷한 대각선을 형성해 '손 → 벽 연결점' 대각선 요구를 정확히 충족한다. 시선 타깃(사슬/고정점)이 행동과 일치한다.",
    "built_space": "암벽 클로즈업과 뒤편 어두운 갱도 입구가 로케이션과 부합한다. 벽 연결점은 프레임 우측 중단에 볼트로 고정된 아이플레이트 1개, 거기서 사슬 1줄이 나온다 — 개수·배치 모두 맞다. 인물이 벽 바로 옆에 밀착해 있어 '벽에 사슬로 묶인 죄수 위치'가 성립하고, 스케일도 과장 없이 실제 크기로 읽힌다. 프레이밍은 요구된 얼굴 클로즈업(얼굴이 중앙·상반 차지, 손과 사슬이 하단 가로지름)에 정확히 부합한다.",
    "entities": "한국인 남성, 검은 머리, 20대 후반으로 읽히는 젊은 얼굴로 캐릭터 레퍼런스에 A보다 가깝다. 의상은 흙 묻은 회백색 셔츠(감금 상태에 부합). 사슬은 굵은 단조 링크로 레퍼런스의 두께·마모감과 유사하다. 깁스는 소매에 가려 보이지 않으나 프레이밍상 감점 대상 아님.",
    "hard_violations": [],
    "physics": "오른손은 벽 고정점 근처 사슬을, 왼손은 그 아래 사슬을 실제로 움켜쥐고 있으며 손등 힘줄과 관절이 하얗게 힘이 들어가 있다. 사슬은 두 손 사이와 고정점까지 처짐 없이 팽팽하다. 몸은 벽에 기대며 뒤로 체중을 실어 당기는 자세로, 지지와 반작용이 명확하다. 공중에 뜬 요소 없음."
   }
  ],
  "totals": {
   "A": 10,
   "B": 17
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "사슬–벽 고정점 대각선은 살렸으나 프레이밍이 클로즈업이 아닌 바스트급이고 얼굴이 중앙 상단이 아닌 좌상단으로 밀렸으며, 인물 연령대가 20대 중반 캐릭터 레퍼런스와 크게 어긋나고 사슬 굵기도 레퍼런스보다 가늘다."
   },
   {
    "label": "B",
    "score": 9,
    "verdict_ko": "얼굴이 프레임 중앙·상반을 차지하고 양손과 팽팽한 사슬이 하단을 가로지르며 우측 벽 연결점으로 대각선을 이루는 지정 구도를 정확히 구현했고, 저조도 동굴 톤·굵은 사슬 재질·젊은 인물 인상 모두 레퍼런스에 부합한다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L37B06.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:787232>"
   },
   {
    "label": "PROP REFERENCE — 굵은 쇠사슬: the exact object appearing in this shot; match its look, material and wear exactly.",
    "path": "<bytes:963202>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "지문에 명시된 깁스(casted arm)가 양팔 어디에도 없습니다.",
     "fix_en": "Cover one of the visible forearms with a white medical cast. Preserve the character's face, hands, clothing, the iron chain, the cave wall, the lighting, and the exact framing.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "레퍼런스의 지정 의상(초록색 코트와 자주색 터틀넥)이 아닌 낡은 베이지색 셔츠를 입고 있습니다.",
     "fix_en": "Change the character's shirt to a dark green wool coat over a maroon turtleneck sweater. Preserve the character's face, hands, the iron chain, the wall anchor, the rock textures, the lighting, and the exact framing.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Cover one of the visible forearms with a white medical cast. Preserve the character's face, hands, clothing, the iron chain, the cave wall, the lighting, and the exact framing.\n- Change the character's shirt to a dark green wool coat over a maroon turtleneck sweater. Preserve the character's face, hands, the iron chain, the wall anchor, the rock textures, the lighting, and the exact framing.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·앵커 위치·사슬 대각선은 충족하지만, 캐릭터 레퍼런스의 녹색 코트/버건디 터틀넥 대신 회색 리넨 셔츠를 입혔고 '부상 팔 깁스 유지' 조항이 누락되어 의상·지속상태 계약에서 실점."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 구도·우측 벽 고정점·팽팽한 사슬 대각선을 유지하면서 캐릭터 레퍼런스 의상(녹색 코트·버건디 터틀넥)과 깁스한 팔이라는 계약 항목을 추가로 충족했고, 계약이 요구하지 않은 변경(얼굴·배경·조명 톤)은 미미한 수준."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.528,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.528,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "프레이밍·앵커 위치·사슬 대각선은 충족하지만, 캐릭터 레퍼런스의 녹색 코트/버건디 터틀넥 대신 회색 리넨 셔츠를 입혔고 '부상 팔 깁스 유지' 조항이 누락되어 의상·지속상태 계약에서 실점."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 클로즈업 구도·우측 벽 고정점·팽팽한 사슬 대각선을 유지하면서 캐릭터 레퍼런스 의상(녹색 코트·버건디 터틀넥)과 깁스한 팔이라는 계약 항목을 추가로 충족했고, 계약이 요구하지 않은 변경(얼굴·배경·조명 톤)은 미미한 수준."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "시선과 얼굴은 프레임 오른쪽 벽의 쇠사슬 고정점 쪽을 향해 있고, 두 손은 모두 사슬을 쥐고 오른쪽 벽 앵커 방향으로 당기는 축과 일치한다. 사슬은 좌하단에서 우상단 앵커로 하드 대각선을 형성해 지시된 방향과 정확히 맞는다. 얼굴은 중앙~상반부, 손과 사슬은 하단부에 진입해 프레임 레이아웃과 일치.",
      "built_space": "동굴 암벽이 화면 오른쪽에 실제 스케일로 서 있고, 벽에 볼트로 박힌 금속 플레이트+링 앵커가 우측 가장자리 중단에 위치. 앵커 개수는 하나로 정확하며 배경 깊이에 어두운 동굴 통로와 희미한 램프 반사광이 보여 레퍼런스 장소와 일치. 배경 요소가 전경으로 과장되지 않았다.",
      "entities": "20대 중반 한국 남성, 검은 머리, 캐릭터 레퍼런스의 이목구비와 부합. 의상은 레퍼런스의 녹색 울 코트+버건디 터틀넥이 그대로 유지되어 있고, 왼팔에는 계약이 요구한 흰색 깁스/붕대가 확인된다. 사슬은 굵고 부식된 단조 링크로 프롭 레퍼런스와 부합(레퍼런스보다 링크가 다소 가늘지만 동종 재질).",
      "hard_violations": [],
      "physics": "몸은 벽 옆에 버티고 서서 상체를 뒤로 젖히며 당기는 자세, 두 손 모두 사슬을 실제로 감싸 쥐었고 힘줄·손등 근육이 긴장으로 부풀어 있다. 사슬은 앵커에서 손까지 팽팽하게 직선으로 당겨져 장력이 물리적으로 성립. 공중에 뜬 물체 없음."
     },
     {
      "label": "B",
      "direction": "동일하게 시선과 얼굴이 우측 앵커 쪽을 향하고 두 손이 사슬을 쥐고 당기는 방향이 앵커와 일치. 사슬 대각선도 좌하단→우상단으로 규정대로 형성. 방향 축은 A와 동등하게 정확.",
      "built_space": "우측 암벽과 볼트 플레이트 앵커 1개, 배경 어두운 동굴 통로와 주황 반사광까지 A와 거의 동일한 공간 구성. 스케일 왜곡 없음.",
      "entities": "동일 인물 얼굴이지만 의상이 캐릭터 레퍼런스의 녹색 코트·버건디 터틀넥에서 회갈색 낡은 리넨 셔츠(수의 같은 옷)로 전면 교체됨. 또한 계약이 지속하라고 명시한 '깁스한 부상 팔'이 사라져 왼팔이 맨 셔츠 소매만 보인다. 사슬 자체는 프롭 레퍼런스와 부합.",
      "hard_violations": [],
      "physics": "상체를 젖히고 두 손으로 사슬을 감싸 쥔 채 당기는 자세, 사슬은 앵커까지 팽팽. 지지·장력은 물리적으로 성립하며 떠 있는 물체 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "클로즈업 프레임·우측 앵커·팽팽한 대각선 사슬을 모두 충족하면서 캐릭터 레퍼런스의 녹색 코트/버건디 터틀넥과 'CARRIED STATE'가 명시한 깁스한 부상 팔까지 유지 — 계약이 요구한 항목을 잃지 않은 쪽."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "구도·방향·물리는 A와 동등하나 계약이 요구하지 않은 의상 전면 교체(녹색 코트→낡은 리넨 셔츠)와 명시적 지속 조건인 '깁스한 부상 팔' 삭제라는 두 가지 손실을 초래해 얻은 것보다 잃은 것이 크다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "클로즈업 프레임·우측 앵커·팽팽한 대각선 사슬을 모두 충족하면서 캐릭터 레퍼런스의 녹색 코트/버건디 터틀넥과 'CARRIED STATE'가 명시한 깁스한 부상 팔까지 유지 — 계약이 요구한 항목을 잃지 않은 쪽."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "구도·방향·물리는 A와 동등하나 계약이 요구하지 않은 의상 전면 교체(녹색 코트→낡은 리넨 셔츠)와 명시적 지속 조건인 '깁스한 부상 팔' 삭제라는 두 가지 손실을 초래해 얻은 것보다 잃은 것이 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "시선과 얼굴은 프레임 오른쪽 벽의 쇠사슬 고정점 쪽을 향해 있고, 두 손은 모두 사슬을 쥐고 오른쪽 벽 앵커 방향으로 당기는 축과 일치한다. 사슬은 좌하단에서 우상단 앵커로 하드 대각선을 형성해 지시된 방향과 정확히 맞는다. 얼굴은 중앙~상반부, 손과 사슬은 하단부에 진입해 프레임 레이아웃과 일치.",
      "built_space": "동굴 암벽이 화면 오른쪽에 실제 스케일로 서 있고, 벽에 볼트로 박힌 금속 플레이트+링 앵커가 우측 가장자리 중단에 위치. 앵커 개수는 하나로 정확하며 배경 깊이에 어두운 동굴 통로와 희미한 램프 반사광이 보여 레퍼런스 장소와 일치. 배경 요소가 전경으로 과장되지 않았다.",
      "entities": "20대 중반 한국 남성, 검은 머리, 캐릭터 레퍼런스의 이목구비와 부합. 의상은 레퍼런스의 녹색 울 코트+버건디 터틀넥이 그대로 유지되어 있고, 왼팔에는 계약이 요구한 흰색 깁스/붕대가 확인된다. 사슬은 굵고 부식된 단조 링크로 프롭 레퍼런스와 부합(레퍼런스보다 링크가 다소 가늘지만 동종 재질).",
      "hard_violations": [],
      "physics": "몸은 벽 옆에 버티고 서서 상체를 뒤로 젖히며 당기는 자세, 두 손 모두 사슬을 실제로 감싸 쥐었고 힘줄·손등 근육이 긴장으로 부풀어 있다. 사슬은 앵커에서 손까지 팽팽하게 직선으로 당겨져 장력이 물리적으로 성립. 공중에 뜬 물체 없음."
     },
     {
      "label": "A",
      "direction": "동일하게 시선과 얼굴이 우측 앵커 쪽을 향하고 두 손이 사슬을 쥐고 당기는 방향이 앵커와 일치. 사슬 대각선도 좌하단→우상단으로 규정대로 형성. 방향 축은 A와 동등하게 정확.",
      "built_space": "우측 암벽과 볼트 플레이트 앵커 1개, 배경 어두운 동굴 통로와 주황 반사광까지 A와 거의 동일한 공간 구성. 스케일 왜곡 없음.",
      "entities": "동일 인물 얼굴이지만 의상이 캐릭터 레퍼런스의 녹색 코트·버건디 터틀넥에서 회갈색 낡은 리넨 셔츠(수의 같은 옷)로 전면 교체됨. 또한 계약이 지속하라고 명시한 '깁스한 부상 팔'이 사라져 왼팔이 맨 셔츠 소매만 보인다. 사슬 자체는 프롭 레퍼런스와 부합.",
      "hard_violations": [],
      "physics": "상체를 젖히고 두 손으로 사슬을 감싸 쥔 채 당기는 자세, 사슬은 앵커까지 팽팽. 지지·장력은 물리적으로 성립하며 떠 있는 물체 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S102sh1__bgfirst_bg.png",
   "bg_asset_id": "3ff55a5d-1ae5-489b-93c4-01f57e4578e9",
   "bg_record_key": "S102sh1::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S103sh4": {
  "input_fingerprint": "12cc668c1eb9281e",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): evening, approaching night.\n\nSHOT TEXT (authoritative, Korean): 극심한 배신감과 회한으로 일그러진 고산들의 두 눈에서 뜨거운 눈물이 왈칵 쏟아지는 얼굴 클로즈업.\n\nLOCATION (lock): The exposed island rock shelf just above the evening surf. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At the eye level of 고산들’s reclined head, the camera holds close and angles slightly across his face rather than squarely confronting it. His contorted face occupies most of the frame, with both eyes retained in focus as tears break free and only a narrow margin of rock preserves his physical position.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 고산들 in the middle-center of the frame, foreground.\n- KEY BACKGROUND ELEMENTS: shore rock beside his head (Supporting his reclined head); used as Leaves a narrow contextual edge around the reclined head.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Dim evening ambience holds a restrained, high-contrast close-up while keeping the tears and seawater visible across his face.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same wet rocks, darkening sea, evening light, soaked clothing, and prone position. Exclude the full-body view from the close-up; focus on the man's tears and betrayal.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Sandel retains the back, chest, and shoulder wounds while lying drenched on the rocks.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 고산들 (Korean man in his mid-40s, closely cropped dark hair, cold angular face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 7,
   "B": 5,
   "C": 9
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 7,
    "verdict_ko": "짧은 머리·젖은 셔츠·암반 접촉 모두 레퍼런스와 부합하고 눈물도 보이나, 좌상단 배경 바위가 전경처럼 크게 걸려 실제 스케일 지시를 약간 침해하고 눈물의 '왈칵 쏟아지는' 감정 강도가 C보다 약함."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "눈물과 젖은 상태는 좋으나 이마를 덮는 젖은 장발이 '아주 짧게 친 머리'라는 캐릭터 레퍼런스와 명백히 어긋나고, 카메라도 누운 머리 눈높이가 아닌 위에서 내려다보는 축이라 카메라 지시에서 감점."
   },
   {
    "label": "C",
    "score": 9,
    "verdict_ko": "누운 머리 눈높이에서 얼굴을 비스듬히 담은 클로즈업, 두 눈 초점 유지와 흘러내리는 눈물, 짧은 머리·젖은 셔츠·저녁 해안 암반 여백까지 카메라·인물·장소 락을 모두 충족."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S103sh2_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 고산들: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:903959>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 샷과 프롬프트에 명시된 가슴과 어깨 부위의 붉은 상처가 화면 우측에 노출된 맨살에 보이지 않고 깨끗하게 묘사되어 있습니다.",
     "fix_en": "Render the required raw, bloody wounds onto the exposed bare skin of his chest and shoulder on the right side of the frame, preserving the man's tearful face, wet hair, shirt, the rock background, the lighting, and the exact framing unchanged.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Render the required raw, bloody wounds onto the exposed bare skin of his chest and shoulder on the right side of the frame, preserving the man's tearful face, wet hair, shirt, the rock background, the lighting, and the exact framing unchanged.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "프레이밍·앵글·눈물·로케이션은 정확하나, 프레임 안에 드러난 어깨·가슴에 상처가 전혀 없어 '등·가슴·어깨 상처를 유지한 채 젖어 누워 있다'는 CARRIED STATE와 이전 스틸의 상처 연속성이 끊긴다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "A와 동일한 얼굴·헤어·의상·배경·조명을 그대로 유지한 채(불필요한 변경 없음) 프레임에 들어온 어깨·가슴의 상처만 복원해 CARRIED STATE와 이전 스틸 연속성을 충족 — 얻은 것은 크고 잃은 것은 없다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.667,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.667,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "프레이밍·앵글·눈물·로케이션은 정확하나, 프레임 안에 드러난 어깨·가슴에 상처가 전혀 없어 '등·가슴·어깨 상처를 유지한 채 젖어 누워 있다'는 CARRIED STATE와 이전 스틸의 상처 연속성이 끊긴다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "A와 동일한 얼굴·헤어·의상·배경·조명을 그대로 유지한 채(불필요한 변경 없음) 프레임에 들어온 어깨·가슴의 상처만 복원해 CARRIED STATE와 이전 스틸 연속성을 충족 — 얻은 것은 크고 잃은 것은 없다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "요구된 눈높이 비스듬 클로즈업·두 눈 유지·눈물, 좁은 바위 여백을 모두 만족하고, CARRIED STATE가 명시한 어깨·가슴 상처까지 프레임 안에서 유지해 이전 스틸과의 연속성이 정확하다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "구도·시선·조명은 A와 거의 동일하나, 계약이 명시한 어깨·가슴 상처가 프레임 안에서 지워져 이전 스틸의 부상 상태 연속성을 잃었다 — 요구되지 않은 손실이므로 감점."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.528
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.528
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "요구된 눈높이 비스듬 클로즈업·두 눈 유지·눈물, 좁은 바위 여백을 모두 만족하고, CARRIED STATE가 명시한 어깨·가슴 상처까지 프레임 안에서 유지해 이전 스틸과의 연속성이 정확하다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "구도·시선·조명은 A와 거의 동일하나, 계약이 명시한 어깨·가슴 상처가 프레임 안에서 지워져 이전 스틸의 부상 상태 연속성을 잃었다 — 요구되지 않은 손실이므로 감점."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 13,
     "B": 17
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S103sh2"
  }
 },
 "S108sh4::bgfirst_bg": {
  "input_fingerprint": "df087d300e6eea89",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 창고 안에서 구창모가 두 손으로 입을 꽉 틀어막은 채 숨을 참는 굳은 얼굴.\n\nLOCATION (lock): The dark storage annex inside the slaughterhouse, behind its small closed side door.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From directly in front of 구창모 at eye level, the dolly-in reaches a tight close frame and settles on his rigid face above both hands sealing his mouth. His face and hands occupy the center while the storage room falls away at the margins; the composition leaves a small amount of lower-frame space for the coming tilt toward his hands.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 구창모 in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 창고 내부 (Dark and quiet); used as Barely resolved spatial boundary behind the close portrait.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key darkness isolates the pale face and compressed hands while the surrounding storage room remains minimally legible.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 어두운 창고 안에서 구창모가 두 손으로 입을 꽉 틀어막은 채 숨을 참는 굳은 얼굴.\n\nLOCATION (lock): The dark storage annex inside the slaughterhouse, behind its small closed side door.\n\nTIME OF DAY (lock): night.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From directly in front of 구창모 at eye level, the dolly-in reaches a tight close frame and settles on his rigid face above both hands sealing his mouth. His face and hands occupy the center while the storage room falls away at the margins; the composition leaves a small amount of lower-frame space for the coming tilt toward his hands.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 구창모 in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 창고 내부 (Dark and quiet); used as Barely resolved spatial boundary behind the close portrait.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key darkness isolates the pale face and compressed hands while the surrounding storage room remains minimally legible.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S108sh4__bgfirst_bg.png",
  "asset_id": "0317598f-6729-4a84-86c5-bc341d2cde55",
  "input_asset_ids": [
   "30852785-121e-4505-bb30-8fd1e291a71a",
   "91ee172a-79c0-44c1-afcb-820865dafad5"
  ]
 },
 "S108sh4": {
  "input_fingerprint": "d1f0e09cd8f9a2d6",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 창고 안에서 구창모가 두 손으로 입을 꽉 틀어막은 채 숨을 참는 굳은 얼굴.\n\nLOCATION (lock): The dark storage annex inside the slaughterhouse, behind its small closed side door. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From directly in front of 구창모 at eye level, the dolly-in reaches a tight close frame and settles on his rigid face above both hands sealing his mouth. His face and hands occupy the center while the storage room falls away at the margins; the composition leaves a small amount of lower-frame space for the coming tilt toward his hands.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 구창모 in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 창고 내부 (Dark and quiet); used as Barely resolved spatial boundary behind the close portrait.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key darkness isolates the pale face and compressed hands while the surrounding storage room remains minimally legible.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (Korean man in his late 40s, short dark hair, lean sun-darkened face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 창고 안에서 구창모가 두 손으로 입을 꽉 틀어막은 채 숨을 참는 굳은 얼굴.\n\nLOCATION (lock): The dark storage annex inside the slaughterhouse, behind its small closed side door. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From directly in front of 구창모 at eye level, the dolly-in reaches a tight close frame and settles on his rigid face above both hands sealing his mouth. His face and hands occupy the center while the storage room falls away at the margins; the composition leaves a small amount of lower-frame space for the coming tilt toward his hands.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 구창모 in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 창고 내부 (Dark and quiet); used as Barely resolved spatial boundary behind the close portrait.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key darkness isolates the pale face and compressed hands while the surrounding storage room remains minimally legible.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (Korean man in his late 40s, short dark hair, lean sun-darkened face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 창고 안에서 구창모가 두 손으로 입을 꽉 틀어막은 채 숨을 참는 굳은 얼굴.\n\nLOCATION (lock): The dark storage annex inside the slaughterhouse, behind its small closed side door. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: From directly in front of 구창모 at eye level, the dolly-in reaches a tight close frame and settles on his rigid face above both hands sealing his mouth. His face and hands occupy the center while the storage room falls away at the margins; the composition leaves a small amount of lower-frame space for the coming tilt toward his hands.\n- FRAMING SCALE: close-up\n- FRAME LAYOUT: 구창모 in the middle-center of the frame, midground.\n- KEY BACKGROUND ELEMENTS: 창고 내부 (Dark and quiet); used as Barely resolved spatial boundary behind the close portrait.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Low-key darkness isolates the pale face and compressed hands while the surrounding storage room remains minimally legible.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 구창모 (Korean man in his late 40s, short dark hair, lean sun-darkened face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S108sh4__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S108sh4.png"
    },
    {
     "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:808030>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L29B05.png"
    },
    {
     "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:808030>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "연기와 로케이션 일치는 좋으나 카메라가 눈높이가 아닌 뚜렷한 로우앵글이고 레퍼런스의 프레이밍(문·천장 모서리 핏자국)을 사실상 복사해 '레퍼런스 구도를 베끼지 말라'는 지시와 배경 최소화 지시를 어겼다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "정면 눈높이 타이트 클로즈업, 중앙 배치, 두 손으로 입을 막고 숨 참는 굳은 얼굴, 배경 창고가 겨우 식별되는 저조도 — 촬영 지시를 가장 정확히 구현했다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.417,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.417,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "연기와 로케이션 일치는 좋으나 카메라가 눈높이가 아닌 뚜렷한 로우앵글이고 레퍼런스의 프레이밍(문·천장 모서리 핏자국)을 사실상 복사해 '레퍼런스 구도를 베끼지 말라'는 지시와 배경 최소화 지시를 어겼다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "정면 눈높이 타이트 클로즈업, 중앙 배치, 두 손으로 입을 막고 숨 참는 굳은 얼굴, 배경 창고가 겨우 식별되는 저조도 — 촬영 지시를 가장 정확히 구현했다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "all_candidates_fail": false,
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "readings": [
     {
      "label": "A",
      "direction": "구창모가 카메라 정면(눈높이)을 향해 서 있고 시선은 렌즈보다 약간 위/정면으로 굳어 있음. 두 손바닥이 입을 완전히 덮고 코 아래를 눌러 봉하는 방향으로 정확히 향함. 카메라 축은 정면 아이레벨로 지정과 일치.",
      "built_space": "배경에 창고 내부 — 왼쪽 뒤 낡은 금속 선반, 중앙 좌측에 닫힌 작은 측면 문(레버 핸들 1개, 위치 정상), 오른쪽 콘크리트 벽에 어두운 흘러내린 얼룩. 문 손잡이 개수·높이 정상, 문은 닫혀 있음. 배경은 초점이 풀려 '겨우 읽히는 공간 경계'로만 기능.",
      "entities": "40대 후반 한국 남성, 짧은 검은 머리, 마르고 볕에 그을린 각진 얼굴 — 캐릭터 레퍼런스와 골격·헤어라인이 잘 맞음. 작업용 어두운 셔츠(먼지·때 묻음)로 도살장 상황에 부합. 다른 인물 없음, 텍스트·워터마크 없음.",
      "hard_violations": [],
      "physics": "인물은 바닥에 선/앉은 상체로 프레임 하단에서 이어지며 어깨·팔이 몸통에 자연스럽게 연결됨. 두 팔은 팔꿈치가 아래에서 올라와 손이 입을 누르는 근육 긴장이 보이고, 손가락 압력으로 볼살이 눌림. 공중에 떠 있는 요소 없음. 손등 힘줄·땀 반사 등 물리적 표현 일관."
     },
     {
      "label": "B",
      "direction": "카메라가 눈높이가 아니라 뚜렷한 로우앵글(아래에서 위로 올려다봄) — 턱 아래와 천장 모서리가 보여 지정된 '정면 아이레벨'과 어긋남. 시선은 렌즈 쪽 정면 아래로 향함. 두 손은 입을 덮고 있으나 오른손이 왼손 위로 겹쳐 실질적으로 한 손 위주로 덮고 다른 손은 코를 눌러 뒤틀린 형태.",
      "built_space": "배경에 레퍼런스 로케이션의 문·레버 핸들·천장 모서리·핏자국이 거의 그대로 재현되나, 레퍼런스의 카메라 앵글(올려다보는 구도)까지 복제해 '레퍼런스 프레이밍을 베끼지 말라'는 지시를 위반. 오른쪽 천장 모서리의 핏자국/조직 덩어리가 실제 거리 대비 확대되어 배경 요소가 전경급 존재감을 가짐.",
      "entities": "40대 후반 한국 남성, 짧고 정돈된 검은 머리, 어두운 셔츠. 얼굴은 레퍼런스보다 다소 통통하고 매끈해 '마르고 그을린 얼굴' 특성이 약함. 다른 인물 없음, 텍스트 없음.",
      "hard_violations": [],
      "physics": "상체가 화면 하단에서 이어지고 팔이 아래에서 올라와 손이 입을 누름 — 지지는 존재. 다만 손 겹침 구조가 어색해 왼손 손가락이 코를 파고드는 형태이고 손목/팔 연결이 다소 평면적. 공중 부양 요소는 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지정된 정면 아이레벨 타이트 클로즈업, 두 손으로 입을 완전히 봉한 굳은 얼굴, 저조도 창고 배경이 경계로만 남는 등 카메라·프레이밍 지시를 가장 정확히 충족."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "얼굴·손 연기는 살아있으나 로케이션 레퍼런스의 올려다보는 앵글을 그대로 복제해 '정면 아이레벨' 지시를 어겼고, 천장 핏자국이 전경급으로 확대돼 실제 스케일 규정도 위반."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지정된 정면 아이레벨 타이트 클로즈업, 두 손으로 입을 완전히 봉한 굳은 얼굴, 저조도 창고 배경이 경계로만 남는 등 카메라·프레이밍 지시를 가장 정확히 충족."
     },
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "얼굴·손 연기는 살아있으나 로케이션 레퍼런스의 올려다보는 앵글을 그대로 복제해 '정면 아이레벨' 지시를 어겼고, 천장 핏자국이 전경급으로 확대돼 실제 스케일 규정도 위반."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "구창모가 카메라 정면(눈높이)을 향해 서 있고 시선은 렌즈보다 약간 위/정면으로 굳어 있음. 두 손바닥이 입을 완전히 덮고 코 아래를 눌러 봉하는 방향으로 정확히 향함. 카메라 축은 정면 아이레벨로 지정과 일치.",
      "built_space": "배경에 창고 내부 — 왼쪽 뒤 낡은 금속 선반, 중앙 좌측에 닫힌 작은 측면 문(레버 핸들 1개, 위치 정상), 오른쪽 콘크리트 벽에 어두운 흘러내린 얼룩. 문 손잡이 개수·높이 정상, 문은 닫혀 있음. 배경은 초점이 풀려 '겨우 읽히는 공간 경계'로만 기능.",
      "entities": "40대 후반 한국 남성, 짧은 검은 머리, 마르고 볕에 그을린 각진 얼굴 — 캐릭터 레퍼런스와 골격·헤어라인이 잘 맞음. 작업용 어두운 셔츠(먼지·때 묻음)로 도살장 상황에 부합. 다른 인물 없음, 텍스트·워터마크 없음.",
      "hard_violations": [],
      "physics": "인물은 바닥에 선/앉은 상체로 프레임 하단에서 이어지며 어깨·팔이 몸통에 자연스럽게 연결됨. 두 팔은 팔꿈치가 아래에서 올라와 손이 입을 누르는 근육 긴장이 보이고, 손가락 압력으로 볼살이 눌림. 공중에 떠 있는 요소 없음. 손등 힘줄·땀 반사 등 물리적 표현 일관."
     },
     {
      "label": "A",
      "direction": "카메라가 눈높이가 아니라 뚜렷한 로우앵글(아래에서 위로 올려다봄) — 턱 아래와 천장 모서리가 보여 지정된 '정면 아이레벨'과 어긋남. 시선은 렌즈 쪽 정면 아래로 향함. 두 손은 입을 덮고 있으나 오른손이 왼손 위로 겹쳐 실질적으로 한 손 위주로 덮고 다른 손은 코를 눌러 뒤틀린 형태.",
      "built_space": "배경에 레퍼런스 로케이션의 문·레버 핸들·천장 모서리·핏자국이 거의 그대로 재현되나, 레퍼런스의 카메라 앵글(올려다보는 구도)까지 복제해 '레퍼런스 프레이밍을 베끼지 말라'는 지시를 위반. 오른쪽 천장 모서리의 핏자국/조직 덩어리가 실제 거리 대비 확대되어 배경 요소가 전경급 존재감을 가짐.",
      "entities": "40대 후반 한국 남성, 짧고 정돈된 검은 머리, 어두운 셔츠. 얼굴은 레퍼런스보다 다소 통통하고 매끈해 '마르고 그을린 얼굴' 특성이 약함. 다른 인물 없음, 텍스트 없음.",
      "hard_violations": [],
      "physics": "상체가 화면 하단에서 이어지고 팔이 아래에서 올라와 손이 입을 누름 — 지지는 존재. 다만 손 겹침 구조가 어색해 왼손 손가락이 코를 파고드는 형태이고 손목/팔 연결이 다소 평면적. 공중 부양 요소는 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 11,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "totals": {
   "A": 11,
   "B": 16
  },
  "selected": "B",
  "ranking": [
   "B",
   "A"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 6,
    "verdict_ko": "연기와 로케이션 일치는 좋으나 카메라가 눈높이가 아닌 뚜렷한 로우앵글이고 레퍼런스의 프레이밍(문·천장 모서리 핏자국)을 사실상 복사해 '레퍼런스 구도를 베끼지 말라'는 지시와 배경 최소화 지시를 어겼다."
   },
   {
    "label": "B",
    "score": 8,
    "verdict_ko": "정면 눈높이 타이트 클로즈업, 중앙 배치, 두 손으로 입을 막고 숨 참는 굳은 얼굴, 배경 창고가 겨우 식별되는 저조도 — 촬영 지시를 가장 정확히 구현했다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L29B05.png"
   },
   {
    "label": "CHARACTER REFERENCE — 구창모: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:808030>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "입을 틀어막은 두 손의 손가락들이 구조 없이 서로 융합되어 해부학적으로 심각하게 엉망입니다.",
     "fix_en": "Redraw the hands covering his mouth to have correct human anatomy, ensuring a clearly distinct left hand and right hand with proper knuckles and distinct fingers that do not fuse or melt into one another. Keep the man's face, his expression, his clothing, the background walls and door, the lighting, and the exact framing unchanged.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the hands covering his mouth to have correct human anatomy, ensuring a clearly distinct left hand and right hand with proper knuckles and distinct fingers that do not fuse or melt into one another. Keep the man's face, his expression, his clothing, the background walls and door, the lighting, and the exact framing unchanged.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "정면 눈높이 클로즈업, 두 손으로 입을 꽉 틀어막은 굳은 얼굴, 저조도 창고 배경 모두 계약대로 충족하며 피부·손·문틀의 재질감이 선명해 사진적 완성도도 우위."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "구도·인물·장소는 A와 사실상 동일하나 전반적 소프트닝과 콘트라스트 저하로 얼굴·손의 질감과 압력감이 뭉개졌고, 계약이 요구하지 않은 이 화질 손실이 순손실이라 A에 뒤진다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.75
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.75
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "정면 눈높이 클로즈업, 두 손으로 입을 꽉 틀어막은 굳은 얼굴, 저조도 창고 배경 모두 계약대로 충족하며 피부·손·문틀의 재질감이 선명해 사진적 완성도도 우위."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "구도·인물·장소는 A와 사실상 동일하나 전반적 소프트닝과 콘트라스트 저하로 얼굴·손의 질감과 압력감이 뭉개졌고, 계약이 요구하지 않은 이 화질 손실이 순손실이라 A에 뒤진다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "정면 눈높이·양손 봉함·창고 문 배경 등 계약을 충족하지만, 크롭이 더 타이트해 '손 쪽으로 틸트할 하단 여백'이 거의 없고 손등·손가락 접합부가 다소 뭉개져 있으며 오른손 형태가 B보다 덜 명료하다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일한 인물·의상·로케이션을 유지한 채 하단 여백과 손의 해부학적 명료도(손가락 마디, 입술 압박 변형)가 개선되어 '숨 참는 굳은 얼굴 + 두 손 봉함' 클로즈업과 후속 틸트 여백 지시를 더 정확히 만족한다 — 바뀐 것은 크롭과 손 디테일뿐, 정체성·소품·배경 손실 없음."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.875,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.875,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "A"
     },
     "agreed": false
    }
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 7,
      "verdict_ko": "정면 눈높이·양손 봉함·창고 문 배경 등 계약을 충족하지만, 크롭이 더 타이트해 '손 쪽으로 틸트할 하단 여백'이 거의 없고 손등·손가락 접합부가 다소 뭉개져 있으며 오른손 형태가 B보다 덜 명료하다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "동일한 인물·의상·로케이션을 유지한 채 하단 여백과 손의 해부학적 명료도(손가락 마디, 입술 압박 변형)가 개선되어 '숨 참는 굳은 얼굴 + 두 손 봉함' 클로즈업과 후속 틸트 여백 지시를 더 정확히 만족한다 — 바뀐 것은 크롭과 손 디테일뿐, 정체성·소품·배경 손실 없음."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 13
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "A",
   "fix_won": false,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S108sh4__bgfirst_bg.png",
   "bg_asset_id": "0317598f-6729-4a84-86c5-bc341d2cde55",
   "bg_record_key": "S108sh4::bgfirst_bg",
   "chain_winner": false,
   "authority": "plate"
  },
  "ref_mode": "플레이트+엔티티 (2택1: 무콘티 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 },
 "S111sh3": {
  "input_fingerprint": "3aab4271a81d2e4d",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): night, clear full moon after eclipse.\n\nSHOT TEXT (authoritative, Korean): 시체 더미 한가운데에 홀로 서서 무표정한 얼굴로 정인우를 빤히 내려다보는 김지후(강수리영)의 전신.\n\nLOCATION (lock): The corpse-strewn center of the forest ritual clearing after the eclipse begins to pass. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: At 정인우’s seated eye level, the camera begins its low dolly-in from the line between him and 김지후(강수리영), keeping her full body above him in the middle distance. 정인우 remains low in the near edge among the bodies while 김지후(강수리영) occupies the upper center, her downward gaze and his half-risen posture forming a physically continuous threat axis.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 정인우 in the lower-left of the frame, foreground, looks toward 김지후(강수리영); 김지후(강수리영) in the upper-center of the frame, midground, looks toward 정인우.\n- KEY BACKGROUND ELEMENTS: 시체 더미 (육혈귀와 인간의 시체들이 쌓여 있음); used as Foreground and middle-distance terrain that forces the low confrontation line; 피가 흐르는 좁은 내 (Flowing through the devastated site); used as Visible traces crossing the ground between the two characters; 핏빛 웅덩이 (Pooled in depressions across the ground); used as Irregular lower-frame obstacles around 정인우.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Clear full-moon illumination reveals the bodies and blood in stark, cold high contrast after the eclipse has ended.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same ritual clearing, monolith, extinguishing chaos, corpses, blood pools, and moonlight after the eclipse clears. Exclude active mass fighting; show the transformed woman alone above the surviving man.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): The dead human and 육혈귀 victims are sprawled and heaped across the ritual ground, with overlapping torsos and slack limbs forming dense piles. Their bodies rest motionless in and around streams and pools of blood.\n\nIMMOBILE BODIES OBEY GRAVITY: a person who is dead or unconscious\nexerts NO muscular effort. Every part of their body — head, torso,\narms, hands, fingers, legs — rests fully on whatever supports it\n(floor, wall, furniture, their own lap) and hangs or slumps with\ngravity. NEVER show any part of an immobile person's body lifted,\nraised, held up in the air, or posed as if presenting something:\nan object in their grip stays clenched in a hand that itself lies\nfallen on a support — the hand does not hold the object up. If the\ncanonical pose leaves a body part unspecified, resolve it as the\nmost gravity-compliant, fully-supported position.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The corpses of Youngran, headless Sandel, the villagers, and human victims remain sprawled and heaped across the blood-flooded ritual ground. Jihu remains golden-eyed, script-covered, and blood-soaked; Inwoo’s arm remains casted.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 김지후(강수리영) (Korean woman in her early 20s, long black hair, pretty youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "totals": {
   "A": 8,
   "B": 4,
   "C": 6
  },
  "selected": "A",
  "ranking": [
   "A",
   "B",
   "C"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "좌하단 정인우–상단 중앙 김지후의 시선축이 실제로 맞물리고, 맑은 보름달·시체 더미·피 웅덩이 등 조명/장소 락을 모두 충족한 유일한 컷."
   },
   {
    "label": "B",
    "score": 4,
    "verdict_ko": "정인우가 고개를 돌려 시선축이 끊기고 달이 부재하며, 좌측 선돌에 문자형 각자가 노출돼 텍스트 금지 규정에 저촉.  ★위반: [claude-opus] 좌측 선돌 표면에 문자/각자 형태의 표기가 드러나 'No text/annotations' 금지에 저촉"
   },
   {
    "label": "C",
    "score": 6,
    "verdict_ko": "구도와 접지는 무난하나 정인우 시선이 김지후를 살짝 빗나가고 보름달이 화면에 없어 조명 락이 미흡, 지정되지 않은 도검 소품도 추가됨.  ★위반: [gpt] 전경 오른쪽에 프롬프트나 기준 장면에 없는 긴 검은 봉·칼 형태의 물체가 추가되어 있다. / [gpt] 좌우 시체 더미의 여러 인물이 팔꿈치와 무릎으로 상체를 세우거나 서로를 붙드는 듯한 근육성 자세를 취해, 모든 사지가 완전히 지지되어야 하는 사망자 고정 조건을 위반한다."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S110sh41_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 김지후(강수리영): the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:807191>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "이전 장면에 있던 인물들이 죽은 시체로 묘사되었으나, 뒷배경 비석들에 기댄 시체들과 좌우측 중경의 시체들이 중력을 거스른 채 무릎을 꿇거나 서 있는 자세로 굳어 있습니다.",
     "fix_en": "Redraw the gravity-defying corpses—specifically the bodies standing vertically against the monoliths in the background, and the ones frozen on their hands and knees in the left and right midground—so they are fully collapsed, limp, and resting flat on the ground. Preserve Jihu, Inwoo, the monoliths, lighting, and overall framing exactly.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "지후의 피부에 문자가 덮여 있어야 하나(script-covered) 전혀 보이지 않습니다.",
     "fix_en": "Draw dark, visible script marks covering Jihu's exposed skin (face, neck, arms, legs), preserving her pose, clothing, the blood, and the rest of the scene completely unchanged.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the gravity-defying corpses—specifically the bodies standing vertically against the monoliths in the background, and the ones frozen on their hands and knees in the left and right midground—so they are fully collapsed, limp, and resting flat on the ground. Preserve Jihu, Inwoo, the monoliths, lighting, and overall framing exactly.\n- Draw dark, visible script marks covering Jihu's exposed skin (face, neck, arms, legs), preserving her pose, clothing, the blood, and the rest of the scene completely unchanged.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "구도·조명·지면 지지 모두 양호하나, CARRIED STATE가 못박은 '경문으로 뒤덮인 지후'가 전혀 구현되지 않았고 시선도 정인우가 아닌 정면에 가깝게 떨어져 핵심 위협축이 약하다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일 구도·장소·인물을 유지하면서 계약이 명시한 '경문으로 뒤덮인' 지후 상태를 표면에 밀착된 실제 문양으로 복원했고, 시선도 좌하단 정인우를 내려다보게 정확히 잡혀 있어 변경 대비 이득이 명확하다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.464,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.464,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "구도·조명·지면 지지 모두 양호하나, CARRIED STATE가 못박은 '경문으로 뒤덮인 지후'가 전혀 구현되지 않았고 시선도 정인우가 아닌 정면에 가깝게 떨어져 핵심 위협축이 약하다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "동일 구도·장소·인물을 유지하면서 계약이 명시한 '경문으로 뒤덮인' 지후 상태를 표면에 밀착된 실제 문양으로 복원했고, 시선도 좌하단 정인우를 내려다보게 정확히 잡혀 있어 변경 대비 이득이 명확하다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "지시된 와이드 구도(좌하단 정인우·상단 중앙 김지후)와 시선 축을 정확히 구현했고, CARRIED STATE가 요구한 '부적 문자로 뒤덮인' 김지후와 깁스, 선돌·핏내·웅덩이 로케이션이 모두 유지되어 계약 충족도가 높다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "구도와 인물 배치는 유지했으나 계약이 명시한 'script-covered' 문양을 지워버렸고 김지후 얼굴선과 좌측 선돌·시체 배치까지 요청 없이 바꿔 얻은 것보다 잃은 것이 크다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 2.0,
      "B": 1.464
     },
     "adjusted": {
      "A": 2.0,
      "B": 1.464
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "A"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "지시된 와이드 구도(좌하단 정인우·상단 중앙 김지후)와 시선 축을 정확히 구현했고, CARRIED STATE가 요구한 '부적 문자로 뒤덮인' 김지후와 깁스, 선돌·핏내·웅덩이 로케이션이 모두 유지되어 계약 충족도가 높다."
     },
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "구도와 인물 배치는 유지했으나 계약이 명시한 'script-covered' 문양을 지워버렸고 김지후 얼굴선과 좌측 선돌·시체 배치까지 요청 없이 바꿔 얻은 것보다 잃은 것이 크다."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 12,
     "B": 16
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S110sh41"
  }
 },
 "S112sh13": {
  "input_fingerprint": "c8b72b0e2b105c49",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day, clear sky.\n\nSHOT TEXT (authoritative, Korean): 반월 식당 앞에 금월도 꼬마 아이들이 일렬로 서서 천진난만한 표정으로 두 사람을 빤히 바라보는 전경.\n\nLOCATION (lock): The frontage of the village restaurant overlooking the island pier, where the surviving children stand together. The shot takes place here — the attached PREVIOUS SHOT STILL shows this exact place.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: The pan settles at eye height into a wide frontal view of the entire row of 금월도 꼬마 아이들, arranged across the middle of the frame before the restaurant. Their bodies remain small against the exterior, and their direct, innocent attention runs past the camera toward 정인우 and 김지후(강수리영) without either adult entering the image.\n- FRAMING SCALE: wide shot\n- KEY BACKGROUND ELEMENTS: 반월 식당 외부 (아이들이 식당 앞에 일렬로 모여 있음) — The restaurant front faces the camera behind the row of children; used as Provides the fixed architectural backdrop and clarifies where the children have gathered; 선착장 쪽 공간 (두 어른의 위치에서 식당 쪽을 바라보는 방향) — The view is oriented from the dock toward the restaurant side; used as Connects the reveal spatially to the dock-side viewpoint.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Bright natural daylight renders the children’s innocent expressions plainly, creating unease through emotional contrast rather than altered illumination.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nTHIS SHOT CONTINUES THE PREVIOUS SHOT: everything the attached still established — the place, its fixed features and wear, each person's clothing and state — persists. Any person in it who cannot move stays PRECISELY as photographed (body, pose, contact points, held objects); only the camera changes.\n\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the same island pier, boat, restaurant frontage, village, sea, and bright daylight. Exclude the couple as the focal subject; show the remaining children lined up before the restaurant watching them.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The surviving children remain lined up before the restaurant. Jihu remains without the destroyed necklace, and Inwoo’s injured arm remains casted.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 어린 한국인 꼬마 아이 (Korean young child, short dark hair, round youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "readings": [
   {
    "label": "A",
    "direction": "아이들 6명이 정면을 향해 일렬로 서 있고 시선은 대체로 카메라(즉 카메라 뒤 두 어른) 쪽을 향한다. 시선 방향은 샷 텍스트('두 사람을 빤히 바라봄')와 일치. 다만 오른쪽 끝 인물은 아이가 아니라 백팩을 멘 성인 여성(참조 스틸의 김지후)으로 보이며, 프롬프트는 두 어른이 화면에 들어오지 않아야 한다고 못박았다.",
    "built_space": "식당 정면이 아이들 뒤로 정면을 향해 서 있고, 기와 지붕·처마 기둥·유리문·'반월 식당' 간판·플라스틱 상자·수조가 배치되어 구조가 일관된다. 오른쪽으로 선착장과 어선이 보여 부두 쪽 시점 연결도 성립. 건물 스케일과 인물 스케일도 실제 비율에 가깝다.",
    "entities": "아이 5명 중 3명은 현대 티셔츠·반바지, 2명은 캐릭터 참조의 삼베 저고리·감색 두건 복장. 캐릭터 참조 얼굴형(짧은 검은 머리, 둥근 얼굴)은 대체로 부합. 그러나 맨 오른쪽 인물은 성인 체격에 배낭·회색 스웨트셔츠로 참조 스틸의 김지후와 동일 인물로 읽혀, '아이만' 조건과 '어른 미등장' 조건을 위반한다. 간판 '반월 식당' 글자는 정상.",
    "hard_violations": [
     "샷 텍스트가 배제하라고 명시한 어른(배낭 멘 김지후로 읽히는 성인 여성)이 오른쪽 끝에 등장 — 프레임에 들어오면 안 되는 인물 추가"
    ],
    "physics": "모든 아이가 두 발로 지면을 딛고 서 있으며 그림자와 접지가 자연스럽다. 공중에 뜬 물체 없음. 팔은 자연스럽게 늘어뜨림."
   },
   {
    "label": "B",
    "direction": "11명이 넘는 인물이 일렬로 서서 모두 카메라를 정면 응시 — 시선 타깃은 맞다. 그러나 중앙 인물은 배낭 멘 성인 여성으로 카메라를 응시하고 있어 어른이 프레임 중앙에 들어와 있다.",
    "built_space": "식당은 프레임 오른쪽으로 밀려 있고 정면이 아니라 비스듬히 서 있으며, 아이들 줄의 절반 이상이 식당이 아닌 바다·부두 배경 앞에 있다. '식당 앞 일렬' 및 '식당 정면이 배경' 지시와 어긋난다. 간판 글자가 '반월 식작답', '밥월 식' 등 왜곡·오탈자 텍스트로 렌더링됨.",
    "entities": "아이들은 모두 현대 복장이며 캐릭터 참조(삼베 한복 아이)와 일치하는 인물이 하나도 없다. 중앙 성인 여성은 참조 스틸의 김지후(배낭·베이지 스웨트셔츠·카고 팬츠)로 명확히 식별되어 '어른 미등장' 위반. 인원수도 과다.",
    "hard_violations": [
     "프레임에 들어오면 안 되는 어른(김지후)이 화면 중앙에 등장",
     "간판에 왜곡·오기된 문자('반월 식작답', '밥월 식') 렌더링 — 텍스트 오류/판독 불가 표기"
    ],
    "physics": "전원이 지면에 서 있고 접지·그림자 정상. 부유 물체 없음."
   },
   {
    "label": "C",
    "direction": "아이 10명이 식당 앞 줄지어 서서 대부분 카메라 쪽(프레임 왼쪽 앞, 즉 부두에 선 두 어른 위치)을 응시한다. 시선 타깃이 샷 텍스트와 부합하며 어른은 화면에 전혀 없다.",
    "built_space": "'반월 식당' 간판이 달린 단층 식당 건물이 아이들 뒤 배경으로 서 있고, 오른쪽에 선착장·어선·바다·먼 섬이 보여 부두 쪽에서 식당을 바라보는 방향이 성립. 다만 카메라가 약간 비스듬한 3/4 각도라 '정면 와이드'보다 사선이며, 줄이 원근으로 물러난다. 건물·인물 스케일 비율은 사실적.",
    "entities": "모든 인물이 어린 한국 아이로 짧은 검은 머리·둥근 얼굴, 캐릭터 참조의 낡은 삼베 옷·짚신 계열 복장과 잘 맞는다(일부는 현대풍 셔츠 혼재). 어른·배낭·성인 체격 인물 없음. 간판 문구는 '반월 식당 / Banwol Restaurant'로 정상 판독되며 유리문 글자 일부는 다소 임의적.",
    "hard_violations": [],
    "physics": "전원이 두 발로 지면을 딛고 서 있으며 그림자가 발밑에 일관되게 떨어진다. 부유하거나 지지 없는 신체 없음."
   }
  ],
  "totals": {
   "A": 5,
   "B": 2,
   "C": 8
  },
  "selected": "C",
  "ranking": [
   "C",
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 5,
    "verdict_ko": "식당 정면 와이드와 아이들 시선은 잘 맞췄지만, 프롬프트가 명시적으로 배제한 어른(배낭 멘 김지후)이 오른쪽 끝에 들어와 하드 위반이다."
   },
   {
    "label": "B",
    "score": 2,
    "verdict_ko": "어른이 프레임 중앙에 서 있고 간판 문자가 '반월 식작답' 등으로 왜곡·오기되었으며, 아이들 줄이 식당 앞이 아닌 부두 앞으로 밀려 배경 지시도 어긋난다."
   },
   {
    "label": "C",
    "score": 8,
    "verdict_ko": "어른 없이 아이들만 반월 식당 앞에 일렬로 서서 부두 쪽 카메라를 빤히 바라보고, 캐릭터 참조 복장·얼굴과 로케이션·주광 톤이 모두 일치 — 정면성이 약간 사선인 점만 감점."
   }
  ],
  "refs": [
   {
    "label": "PREVIOUS SHOT STILL — a visually related earlier shot of this same place: the location's look, materials, fixed features, lighting mood and each person's clothing are LOCKED to this photo; never copy its camera framing. If this photo shows a character who cannot move (dead or unconscious), that character's exact pose, position and orientation are ALSO LOCKED — treat their body as a fixed prop of the set that only the camera moves around.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S112sh8_sel.png"
   },
   {
    "label": "CHARACTER REFERENCE — 어린 한국인 꼬마 아이: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:904057>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "식당 간판과 유리창에 한글 및 영문 텍스트가 명확하게 렌더링되어 있습니다. 프롬프트는 화면 내 어떠한 텍스트나 글자도 포함되지 않도록 엄격히 금지하고 있습니다.",
     "fix_en": "Remove all text and lettering from the restaurant's main signboard and window glass, leaving only bare weathered panels and plain dusty glass in those spaces. Strictly preserve the current camera framing, the lighting, the building's structure, the background boat and pier, and every child exactly as they are currently posed and dressed.",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 3,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all text and lettering from the restaurant's main signboard and window glass, leaving only bare weathered panels and plain dusty glass in those spaces. Strictly preserve the current camera framing, the lighting, the building's structure, the background boat and pier, and every child exactly as they are currently posed and dressed.\n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "눈높이 정면 와이드·식당 앞 일렬·어른 미등장·시선의 카메라 통과라는 핵심 스테이징을 모두 충족하고 이전 스틸의 부두·어선·햇빛도 이어받았다. 간판/유리창의 상호 글자가 '텍스트 금지' 조항과 마찰하고 일부 글자가 비문이라는 감점이 있으나, 그것이 곧 로케이션(반월 식당) 식별자이기도 하다.  ★위반: [gpt] 옥상 간판과 창문에 다수의 한글 및 영문이 선명하게 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "텍스트 제거라는 이득은 있으나 계약이 요구하지 않은 손실이 크다 — 간판이 아무 글자 없는 빈 판으로 변해 '반월 식당' 로케이션 지시가 사라졌고, 왼쪽 아이의 저고리 여밈·신발 등 의상 디테일과 얼굴 톤까지 함께 바뀌었다. 변경 범위가 넓고 배경 정체성을 훼손해 A보다 뒤진다."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.375,
      "B": 1.857
     },
     "adjusted": {
      "A": 1.125,
      "B": 1.857
     },
     "violations": {
      "A": [
       "[gpt] 옥상 간판과 창문에 다수의 한글 및 영문이 선명하게 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
      ]
     },
     "per_model_winner": {
      "claude-opus": "A",
      "gpt": "B"
     },
     "agreed": false
    }
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 7,
      "verdict_ko": "눈높이 정면 와이드·식당 앞 일렬·어른 미등장·시선의 카메라 통과라는 핵심 스테이징을 모두 충족하고 이전 스틸의 부두·어선·햇빛도 이어받았다. 간판/유리창의 상호 글자가 '텍스트 금지' 조항과 마찰하고 일부 글자가 비문이라는 감점이 있으나, 그것이 곧 로케이션(반월 식당) 식별자이기도 하다.  ★위반: [gpt] 옥상 간판과 창문에 다수의 한글 및 영문이 선명하게 노출되어 ‘어디에도 텍스트 금지’ 조건을 위반한다."
     },
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "텍스트 제거라는 이득은 있으나 계약이 요구하지 않은 손실이 크다 — 간판이 아무 글자 없는 빈 판으로 변해 '반월 식당' 로케이션 지시가 사라졌고, 왼쪽 아이의 저고리 여밈·신발 등 의상 디테일과 얼굴 톤까지 함께 바뀌었다. 변경 범위가 넓고 배경 정체성을 훼손해 A보다 뒤진다."
     }
    ],
    "all_candidates_fail": false
   },
   "reverse_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "아이들 10명이 식당 정면 앞에 일렬로 서서, 시선이 모두 카메라 축을 약간 좌측(프레임 왼쪽 아래)으로 스쳐 지나가는 방향을 응시한다. 즉 카메라를 정면으로 뚫어보기보다 카메라 옆의 두 어른 위치를 바라보는 구도로 읽힌다. 어른(정인우·김지후)은 화면에 등장하지 않아 지시 충족. 카메라는 선착장 쪽에서 식당 쪽을 바라보는 방향, 눈높이 와이드.",
      "built_space": "반월 식당 외관: 좌측에 방충망 미닫이문, 중앙에 알루미늄 새시 유리창, 우측에 작은 차양과 실외기, 슬레이트/함석 지붕과 처마. 지붕 위 간판 프레임은 있으나 글자가 없는 빈 백색판. 우측 배경에 선착장과 어선, 방파제, 원경 섬 능선 — 이전 스틸의 재질·색·광선과 일관. 배경 요소들이 실제 크기 비율로 작게 자리함.",
      "entities": "아이 10명 모두 한국인 어린이, 짧은 검은 머리, 둥근 얼굴 — 캐릭터 레퍼런스 프로필과 부합. 맨 왼쪽 아이는 레퍼런스의 삼베 저고리·허리끈 복장에 가깝고 나머지는 현대 시골 아이 옷차림(때 묻은 니트·면바지·운동화/짚신 혼재). 어른은 없음. 목걸이·깁스 등은 프레임에 어른이 없으므로 해당 없음.",
      "hard_violations": [],
      "physics": "모든 아이가 두 발로 지면을 딛고 서 있으며 그림자가 발 아래 왼쪽으로 짧게 떨어져 정오 무렵 맑은 날 광선과 일치. 공중에 뜬 신체·물체 없음. 지면 접지선과 원근이 자연스럽고 뒤쪽 아이일수록 작아지는 스케일 유지."
     },
     {
      "label": "B",
      "direction": "A와 동일한 아이들 열, 동일한 시선 방향(카메라 좌측을 스쳐 지나감). 어른 부재도 동일. 카메라 축·눈높이 동일.",
      "built_space": "A와 동일한 식당 구조지만 지붕 간판에 '반월 식당 / Banwol Restaurant' 대형 글자, 유리창에도 '반월식당', '잠', '락기학기' 등 붉고 초록인 글자들이 추가됨. 창문 글자는 유리 곡면·반사와 무관하게 평면 스티커처럼 얹혀 있고 '락기학기'는 의미 없는 깨진 한글. 나머지 선착장·배·바다는 A와 동일.",
      "entities": "아이 10명 동일 인물·의상, 한국인 어린이 프로필 부합. 배경 텍스트가 추가된 것 외 인물 차이 없음.",
      "hard_violations": [
       "프롬프트가 명시적으로 금지한 텍스트/캡션류가 화면 전체에 삽입됨(간판·유리창 문구, 'Banwol Restaurant' 영문 포함) — 'No text, captions, watermarks or annotations anywhere' 위반",
       "유리창 글자가 표면 재질감·반사를 따르지 않고 평면 그래픽 오버레이처럼 붙어 있어 MATERIAL REALISM 위반, '락기학기' 등 왜곡된 문자 누출"
      ],
      "physics": "아이들은 모두 두 발로 접지, 그림자 방향 일관, 공중 부양 없음 — A와 동일하게 문제 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "눈높이 와이드·선착장→식당 방향·어른 부재·일렬 아이들 전경을 모두 충족하고, 접지와 스케일도 정확하며 '어떤 텍스트도 금지' 조항을 지켜 간판을 빈 판으로 남긴 점이 결정적이다."
     },
     {
      "label": "B",
      "score": 3,
      "verdict_ko": "인물·구도·광선은 A와 사실상 동일하나, 계약이 요구하지 않은 간판·유리창 문구(영문 포함, '락기학기' 같은 깨진 글자)를 추가해 '텍스트 일절 금지'와 재질 사실성 조항을 정면 위반했다 — 얻은 것은 식당 이름 명시뿐이고 잃은 것이 더 크다."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "눈높이 와이드·선착장→식당 방향·어른 부재·일렬 아이들 전경을 모두 충족하고, 접지와 스케일도 정확하며 '어떤 텍스트도 금지' 조항을 지켜 간판을 빈 판으로 남긴 점이 결정적이다."
     },
     {
      "label": "A",
      "score": 3,
      "verdict_ko": "인물·구도·광선은 A와 사실상 동일하나, 계약이 요구하지 않은 간판·유리창 문구(영문 포함, '락기학기' 같은 깨진 글자)를 추가해 '텍스트 일절 금지'와 재질 사실성 조항을 정면 위반했다 — 얻은 것은 식당 이름 명시뿐이고 잃은 것이 더 크다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "아이들 10명이 식당 정면 앞에 일렬로 서서, 시선이 모두 카메라 축을 약간 좌측(프레임 왼쪽 아래)으로 스쳐 지나가는 방향을 응시한다. 즉 카메라를 정면으로 뚫어보기보다 카메라 옆의 두 어른 위치를 바라보는 구도로 읽힌다. 어른(정인우·김지후)은 화면에 등장하지 않아 지시 충족. 카메라는 선착장 쪽에서 식당 쪽을 바라보는 방향, 눈높이 와이드.",
      "built_space": "반월 식당 외관: 좌측에 방충망 미닫이문, 중앙에 알루미늄 새시 유리창, 우측에 작은 차양과 실외기, 슬레이트/함석 지붕과 처마. 지붕 위 간판 프레임은 있으나 글자가 없는 빈 백색판. 우측 배경에 선착장과 어선, 방파제, 원경 섬 능선 — 이전 스틸의 재질·색·광선과 일관. 배경 요소들이 실제 크기 비율로 작게 자리함.",
      "entities": "아이 10명 모두 한국인 어린이, 짧은 검은 머리, 둥근 얼굴 — 캐릭터 레퍼런스 프로필과 부합. 맨 왼쪽 아이는 레퍼런스의 삼베 저고리·허리끈 복장에 가깝고 나머지는 현대 시골 아이 옷차림(때 묻은 니트·면바지·운동화/짚신 혼재). 어른은 없음. 목걸이·깁스 등은 프레임에 어른이 없으므로 해당 없음.",
      "hard_violations": [],
      "physics": "모든 아이가 두 발로 지면을 딛고 서 있으며 그림자가 발 아래 왼쪽으로 짧게 떨어져 정오 무렵 맑은 날 광선과 일치. 공중에 뜬 신체·물체 없음. 지면 접지선과 원근이 자연스럽고 뒤쪽 아이일수록 작아지는 스케일 유지."
     },
     {
      "label": "A",
      "direction": "A와 동일한 아이들 열, 동일한 시선 방향(카메라 좌측을 스쳐 지나감). 어른 부재도 동일. 카메라 축·눈높이 동일.",
      "built_space": "A와 동일한 식당 구조지만 지붕 간판에 '반월 식당 / Banwol Restaurant' 대형 글자, 유리창에도 '반월식당', '잠', '락기학기' 등 붉고 초록인 글자들이 추가됨. 창문 글자는 유리 곡면·반사와 무관하게 평면 스티커처럼 얹혀 있고 '락기학기'는 의미 없는 깨진 한글. 나머지 선착장·배·바다는 A와 동일.",
      "entities": "아이 10명 동일 인물·의상, 한국인 어린이 프로필 부합. 배경 텍스트가 추가된 것 외 인물 차이 없음.",
      "hard_violations": [
       "프롬프트가 명시적으로 금지한 텍스트/캡션류가 화면 전체에 삽입됨(간판·유리창 문구, 'Banwol Restaurant' 영문 포함) — 'No text, captions, watermarks or annotations anywhere' 위반",
       "유리창 글자가 표면 재질감·반사를 따르지 않고 평면 그래픽 오버레이처럼 붙어 있어 MATERIAL REALISM 위반, '락기학기' 등 왜곡된 문자 누출"
      ],
      "physics": "아이들은 모두 두 발로 접지, 그림자 방향 일관, 공중 부양 없음 — A와 동일하게 문제 없음."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 10,
     "B": 14
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "ref_mode": "prev+엔티티",
  "share_plan": {
   "ref_plan": "prev",
   "prev_anchor_tag": "S112sh8"
  },
  "lane_policy": "ab_select_bypass:prev"
 },
 "S113sh2::bgfirst_bg": {
  "input_fingerprint": "83b9ba99bd666b97",
  "prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 낚시 어선 갑판 위에서 금월도 방향으로 상체를 튼 채 애틋한 표정으로 시선을 고정한 정인우의 전신.\n\nLOCATION (lock): The open deck of the fishing boat just beyond the island harbor, facing back toward the receding pier.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: A waist-height tracking camera remains behind 정인우 on the moving deck, holding his full body in the middle-right as his torso twists back toward 금월도. The receding island is retained across the opposite background, allowing his fixed farewell gaze and the growing physical separation to read within one composition.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 정인우 in the middle-right of the frame, midground, looks toward receding Geumwoldo; receding Geumwoldo in the upper-left of the frame, background.\n- KEY BACKGROUND ELEMENTS: 낚시 어선 갑판 (금월도를 벗어나 육지로 향하는 중) — The deck extends away from the camera toward the boat’s direction of travel; used as Moving platform beneath 정인우 and foreground evidence of departure; 금월도 (배에서 점점 멀어지고 있음); used as Primary distant anchor of 정인우’s backward gaze; 바다와 물결 (물결이 반짝임); used as Separates the boat from the island and carries the reflected daylight.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight over the sparkling water keeps the departure outwardly serene while the restrained contrast holds 정인우’s grief.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "effective_prompt": "Create the EMPTY BACKGROUND PLATE for one film shot — NO PEOPLE, no figures, no body parts, no sketch lines, no arrows anywhere.\n\"Empty\" means no people only: KEEP the location's inherent occupants and stock that define the place — animals in an animal shelter, pen or farm, goods in a market, moored boats in a harbour — unless the shot text explicitly removes them.\nThe FIRST attached image is a thin-line storyboard sketch: use ONLY its camera angle, horizon, perspective and the placement/size of buildings and set masses — ignore the sketched people and arrows entirely. The SECOND attached image (LOCATION PHOTOGRAPH) is the real place: take its architecture, materials, signage and fixed features, and RE-PROJECT them into the sketch's camera. If the photograph's camera differs from the sketch's, the sketch's camera wins.\nHUMAN-SCALE CALIBRATION: derive every structure's true size from human-scale elements — a door ≈ 2m, a window ≈ 1–1.5m wide, one storey ≈ 2.5–3m; never inflate a small structure or shrink a large one.\n\nSHOT TEXT this background must serve (Korean): 낚시 어선 갑판 위에서 금월도 방향으로 상체를 튼 채 애틋한 표정으로 시선을 고정한 정인우의 전신.\n\nLOCATION (lock): The open deck of the fishing boat just beyond the island harbor, facing back toward the receding pier.\n\nTIME OF DAY (lock): day.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: A waist-height tracking camera remains behind 정인우 on the moving deck, holding his full body in the middle-right as his torso twists back toward 금월도. The receding island is retained across the opposite background, allowing his fixed farewell gaze and the growing physical separation to read within one composition.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 정인우 in the middle-right of the frame, midground, looks toward receding Geumwoldo; receding Geumwoldo in the upper-left of the frame, background.\n- KEY BACKGROUND ELEMENTS: 낚시 어선 갑판 (금월도를 벗어나 육지로 향하는 중) — The deck extends away from the camera toward the boat’s direction of travel; used as Moving platform beneath 정인우 and foreground evidence of departure; 금월도 (배에서 점점 멀어지고 있음); used as Primary distant anchor of 정인우’s backward gaze; 바다와 물결 (물결이 반짝임); used as Separates the boat from the island and carries the reflected daylight.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight over the sparkling water keeps the departure outwardly serene while the restrained contrast holds 정인우’s grief.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nRender ONE photorealistic empty location photograph, 16:9, that this shot can be staged inside later. No text or watermarks.",
  "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S113sh2__bgfirst_bg.png",
  "asset_id": "0d368f14-ed64-4cfd-b5dc-5935e7a983ac",
  "input_asset_ids": [
   "8b86a8fe-3549-4cd0-a8ed-9fce4f573257",
   "5b0bd345-0f69-4e6f-85c2-b1bd374e516e"
  ]
 },
 "S113sh2": {
  "input_fingerprint": "2e278d0c4fbe9f98",
  "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 낚시 어선 갑판 위에서 금월도 방향으로 상체를 튼 채 애틋한 표정으로 시선을 고정한 정인우의 전신.\n\nLOCATION (lock): The open deck of the fishing boat just beyond the island harbor, facing back toward the receding pier. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: A waist-height tracking camera remains behind 정인우 on the moving deck, holding his full body in the middle-right as his torso twists back toward 금월도. The receding island is retained across the opposite background, allowing his fixed farewell gaze and the growing physical separation to read within one composition.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 정인우 in the middle-right of the frame, midground, looks toward receding Geumwoldo; receding Geumwoldo in the upper-left of the frame, background.\n- KEY BACKGROUND ELEMENTS: 낚시 어선 갑판 (금월도를 벗어나 육지로 향하는 중) — The deck extends away from the camera toward the boat’s direction of travel; used as Moving platform beneath 정인우 and foreground evidence of departure; 금월도 (배에서 점점 멀어지고 있음); used as Primary distant anchor of 정인우’s backward gaze; 바다와 물결 (물결이 반짝임); used as Separates the boat from the island and carries the reflected daylight.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight over the sparkling water keeps the departure outwardly serene while the restrained contrast holds 정인우’s grief.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains immobilized in its cast as the fishing boat departs.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
  "roll_prompts": {
   "A": "Stage the shot. The FIRST attached image (SHOT BACKGROUND) is the finished empty background of this shot — keep it EXACTLY: its camera, perspective, architecture, lighting and every fixed feature stay untouched. The SECOND attached image (LAYOUT SKETCH) tells you ONLY where the people go: each sketched person's position, screen size, pose and the gaze/motion arrows. Ignore the sketch's background lines. The CHARACTER REFERENCE photographs show the real people.\nPlace the real people into the background at exactly the sketched positions, sizes and poses, following the arrow directions. No sketch lines or arrows may remain.\n\nCreate ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 낚시 어선 갑판 위에서 금월도 방향으로 상체를 튼 채 애틋한 표정으로 시선을 고정한 정인우의 전신.\n\nLOCATION (lock): The open deck of the fishing boat just beyond the island harbor, facing back toward the receding pier. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: A waist-height tracking camera remains behind 정인우 on the moving deck, holding his full body in the middle-right as his torso twists back toward 금월도. The receding island is retained across the opposite background, allowing his fixed farewell gaze and the growing physical separation to read within one composition.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 정인우 in the middle-right of the frame, midground, looks toward receding Geumwoldo; receding Geumwoldo in the upper-left of the frame, background.\n- KEY BACKGROUND ELEMENTS: 낚시 어선 갑판 (금월도를 벗어나 육지로 향하는 중) — The deck extends away from the camera toward the boat’s direction of travel; used as Moving platform beneath 정인우 and foreground evidence of departure; 금월도 (배에서 점점 멀어지고 있음); used as Primary distant anchor of 정인우’s backward gaze; 바다와 물결 (물결이 반짝임); used as Separates the boat from the island and carries the reflected daylight.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight over the sparkling water keeps the departure outwardly serene while the restrained contrast holds 정인우’s grief.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains immobilized in its cast as the fishing boat departs.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere.",
   "B": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea; historical sequences in 1670 Joseon and early-20th-century Korea under Japanese rule; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 낚시 어선 갑판 위에서 금월도 방향으로 상체를 튼 채 애틋한 표정으로 시선을 고정한 정인우의 전신.\n\nLOCATION (lock): The open deck of the fishing boat just beyond the island harbor, facing back toward the receding pier. The shot takes place here — the attached LOCATION PHOTOGRAPH shows the exact spot.\n\nCAMERA & FRAME (follow exactly — this is the composition authority for this still; the reference images supply identity and place, never the framing):\n- CAMERA: A waist-height tracking camera remains behind 정인우 on the moving deck, holding his full body in the middle-right as his torso twists back toward 금월도. The receding island is retained across the opposite background, allowing his fixed farewell gaze and the growing physical separation to read within one composition.\n- FRAMING SCALE: wide shot\n- FRAME LAYOUT: 정인우 in the middle-right of the frame, midground, looks toward receding Geumwoldo; receding Geumwoldo in the upper-left of the frame, background.\n- KEY BACKGROUND ELEMENTS: 낚시 어선 갑판 (금월도를 벗어나 육지로 향하는 중) — The deck extends away from the camera toward the boat’s direction of travel; used as Moving platform beneath 정인우 and foreground evidence of departure; 금월도 (배에서 점점 멀어지고 있음); used as Primary distant anchor of 정인우’s backward gaze; 바다와 물결 (물결이 반짝임); used as Separates the boat from the island and carries the reflected daylight.\nCompose the frame exactly as specified above — camera angle, subject scale and screen placement. Keep true physical scale between people and background elements: fixtures and distant objects occupy only the small screen area their real size and distance dictate; never enlarge a background object into a foreground presence.\n\nLIGHTING & MOOD (these govern light, color and surface rendering):\n- LIGHTING & MOOD: Natural daylight over the sparkling water keeps the departure outwardly serene while the restrained contrast holds 정인우’s grief.\n- MATERIAL REALISM: every object and surface must read as a real,\n  physical material — correct texture, weight, wear and light response\n  (metal reflects, fabric drapes and creases, liquid is glossy and\n  pools, painted or drawn marks sit ON a surface and follow its\n  curvature and lighting). Nothing may look like a flat sticker, a\n  doodle or a graphic overlay pasted onto the frame.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract, immobility contract or shot-text\ndirection says otherwise, people read as alive in mid-moment —\nbelievable weight shift, hands engaged with something real, gaze on\nthe target the SHOT TEXT implies; avoid a stiff attention stance (feet\ntogether, arms hanging straight down) and a blank stare into the lens.\nIf the SHOT TEXT or any POSE/IMMOBILE contract above stages stillness,\ndeath, sleep, unconsciousness, restraint, drill or an explicit\ndirect-to-camera look, follow THAT exactly — this clause never\noverrides it.\n\nDRAWN MARKS KEEP THEIR SHAPE: any mark the text describes as drawn,\npainted or traced (a circle, a line, a symbol) is a STROKE sitting on\nthe surface — an outline whose interior still shows the underlying\nsurface (wall, skin, paper). Render its stated shape faithfully: a\ndrawn circle stays an open ring of brush-width, never filled into a\nsolid disc, unless the text explicitly says it is filled.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Inwoo’s injured arm remains immobilized in its cast as the fishing boat departs.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 정인우 (Korean man in his mid-20s, dark hair, handsome youthful face) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text names — its sex, age, build, skin and grooming must unmistakably match that person's profile above.\n\nNo text, captions, watermarks or annotations anywhere."
  },
  "roll_refs": {
   "A": [
    {
     "label": "SHOT BACKGROUND",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S113sh2__bgfirst_bg.png"
    },
    {
     "label": "LAYOUT SKETCH (people placement only)",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/conti/conti_S113sh2.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:787232>"
    }
   ],
   "B": [
    {
     "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
     "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L15B07.png"
    },
    {
     "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
     "path": "<bytes:787232>"
    }
   ]
  },
  "judge_flip": {
   "forward_raw": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "readings": [
     {
      "label": "A",
      "direction": "정인우는 프레임 중앙(약간 왼쪽~중앙)에 서서 상체와 시선을 왼쪽 뒤편, 즉 배 항적 너머 멀어지는 섬/부두 방향으로 고정하고 있다. 시선은 화면 좌상단의 섬·방파제 라인에 정확히 착지하며, 카메라는 그의 뒤·측면에서 허리 높이로 잡아 등과 옆얼굴이 보인다. 애틋한 감정이 눈꺼풀과 다문 입에서 읽힌다.",
      "built_space": "어선 갑판: 조타실은 화면 오른쪽, 조타실 유리창 안에 조향 휠 1개만 보이고 계기판·레버가 위치 그대로다. 좌현 난간, 청·백 도장 불워크, 젖은 갑판 등 로케이션 레퍼런스와 일치. 조타실 문·환기 그릴·마스트 등 부속물 개수와 배치가 참조와 맞다. 사람이 앉는 자리는 없고 갑판 위에 서 있어 무리 없다.",
      "entities": "인물은 20대 중반 한국 남성, 검은 단발머리, 참조 얼굴과 상당히 유사. 왼팔은 하늘색 팔걸이 슬링에 걸쳐 고정되어 있으나 '캐스트(깁스)'라기보다 슬링만 보인다. 배경의 섬·방파제·붉은 크레인·건물이 로케이션 사진과 동일. 바다에 항적 거품이 있어 이동 중임이 읽힌다. 텍스트·워터마크 없음.",
      "hard_violations": [],
      "physics": "두 발이 젖은 갑판에 확실히 닿아 체중을 지지하며, 오른팔은 자연스레 내려오고 왼팔은 슬링이 지탱한다. 공중에 뜬 물체 없음. 배의 항적이 진행 방향과 일치해 이동 플랫폼으로 성립. 그림자와 젖은 갑판 반사가 물리적으로 타당."
     },
     {
      "label": "B",
      "direction": "정인우는 프레임 중앙에서 걷는 자세로 몸통은 오른쪽(선수 방향)으로 향하고 고개만 왼쪽 뒤로 돌려 섬 쪽을 본다. 시선은 대체로 화면 좌측 섬 방향으로 향하나 다소 카메라·중경 쪽으로 흐르고, 시선 라인이 상단 좌측의 부두에 정확히 꽂히기보다 애매하다. 카메라는 측면에 가깝고 그의 뒤가 아니라 옆에서 잡아 '뒤에서 따라가는' 지시와 어긋난다.",
      "built_space": "조타실 오른쪽, 조향 휠 1개, 계기판·레버 정상. 좌현 난간과 불워크 배치 일치. 다만 인물이 조타실 문 바로 앞을 가리며 문 손잡이 부분이 어색하게 겹친다. 해수면에 항적 거품이 없어 정박에 가까운 잔잔한 수면 — '이동 중' 증거가 약하다.",
      "entities": "20대 남성, 검은 머리로 참조와 유사하나 얼굴형·이목구비가 레퍼런스와 다소 차이. 오른팔에 흰 붕대/깁스가 감겨 있어 캐스트 조건은 충족. 배경 섬·붉은 크레인·부두는 로케이션과 일치. 텍스트 없음.",
      "hard_violations": [],
      "physics": "양발이 갑판에 닿아 있고 걷는 중간 자세로 무게 이동이 보이나, 팔은 깁스한 채 몸 앞으로 어색하게 접혀 있고 왼손 위치가 불분명하게 조타실 창에 겹친다. 부유하는 물체는 없다. 다만 배가 움직인다는 물리적 증거(항적)가 부재."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "허리 높이·뒤쪽 카메라, 좌상단 멀어지는 섬으로 고정된 애틋한 시선, 항적으로 드러나는 이동 중 갑판까지 지시를 충실히 구현했다. 다만 인물이 지정된 '중앙 우측'보다 약간 왼쪽이고 깁스가 슬링으로 대체된 점이 감점."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "깁스는 명확하나 카메라가 뒤가 아닌 측면이고, 잔잔한 수면에 항적이 없어 '멀어지는 중'이라는 이동 증거가 사라졌으며 시선도 섬에 정확히 고정되지 않아 핵심 스테이징에서 밀린다."
     }
    ],
    "all_candidates_fail": false
   },
   "forward_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "허리 높이·뒤쪽 카메라, 좌상단 멀어지는 섬으로 고정된 애틋한 시선, 항적으로 드러나는 이동 중 갑판까지 지시를 충실히 구현했다. 다만 인물이 지정된 '중앙 우측'보다 약간 왼쪽이고 깁스가 슬링으로 대체된 점이 감점."
     },
     {
      "label": "B",
      "score": 5,
      "verdict_ko": "깁스는 명확하나 카메라가 뒤가 아닌 측면이고, 잔잔한 수면에 항적이 없어 '멀어지는 중'이라는 이동 증거가 사라졌으며 시선도 섬에 정확히 고정되지 않아 핵심 스테이징에서 밀린다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "정인우는 프레임 중앙(약간 왼쪽~중앙)에 서서 상체와 시선을 왼쪽 뒤편, 즉 배 항적 너머 멀어지는 섬/부두 방향으로 고정하고 있다. 시선은 화면 좌상단의 섬·방파제 라인에 정확히 착지하며, 카메라는 그의 뒤·측면에서 허리 높이로 잡아 등과 옆얼굴이 보인다. 애틋한 감정이 눈꺼풀과 다문 입에서 읽힌다.",
      "built_space": "어선 갑판: 조타실은 화면 오른쪽, 조타실 유리창 안에 조향 휠 1개만 보이고 계기판·레버가 위치 그대로다. 좌현 난간, 청·백 도장 불워크, 젖은 갑판 등 로케이션 레퍼런스와 일치. 조타실 문·환기 그릴·마스트 등 부속물 개수와 배치가 참조와 맞다. 사람이 앉는 자리는 없고 갑판 위에 서 있어 무리 없다.",
      "entities": "인물은 20대 중반 한국 남성, 검은 단발머리, 참조 얼굴과 상당히 유사. 왼팔은 하늘색 팔걸이 슬링에 걸쳐 고정되어 있으나 '캐스트(깁스)'라기보다 슬링만 보인다. 배경의 섬·방파제·붉은 크레인·건물이 로케이션 사진과 동일. 바다에 항적 거품이 있어 이동 중임이 읽힌다. 텍스트·워터마크 없음.",
      "hard_violations": [],
      "physics": "두 발이 젖은 갑판에 확실히 닿아 체중을 지지하며, 오른팔은 자연스레 내려오고 왼팔은 슬링이 지탱한다. 공중에 뜬 물체 없음. 배의 항적이 진행 방향과 일치해 이동 플랫폼으로 성립. 그림자와 젖은 갑판 반사가 물리적으로 타당."
     },
     {
      "label": "B",
      "direction": "정인우는 프레임 중앙에서 걷는 자세로 몸통은 오른쪽(선수 방향)으로 향하고 고개만 왼쪽 뒤로 돌려 섬 쪽을 본다. 시선은 대체로 화면 좌측 섬 방향으로 향하나 다소 카메라·중경 쪽으로 흐르고, 시선 라인이 상단 좌측의 부두에 정확히 꽂히기보다 애매하다. 카메라는 측면에 가깝고 그의 뒤가 아니라 옆에서 잡아 '뒤에서 따라가는' 지시와 어긋난다.",
      "built_space": "조타실 오른쪽, 조향 휠 1개, 계기판·레버 정상. 좌현 난간과 불워크 배치 일치. 다만 인물이 조타실 문 바로 앞을 가리며 문 손잡이 부분이 어색하게 겹친다. 해수면에 항적 거품이 없어 정박에 가까운 잔잔한 수면 — '이동 중' 증거가 약하다.",
      "entities": "20대 남성, 검은 머리로 참조와 유사하나 얼굴형·이목구비가 레퍼런스와 다소 차이. 오른팔에 흰 붕대/깁스가 감겨 있어 캐스트 조건은 충족. 배경 섬·붉은 크레인·부두는 로케이션과 일치. 텍스트 없음.",
      "hard_violations": [],
      "physics": "양발이 갑판에 닿아 있고 걷는 중간 자세로 무게 이동이 보이나, 팔은 깁스한 채 몸 앞으로 어색하게 접혀 있고 왼손 위치가 불분명하게 조타실 창에 겹친다. 부유하는 물체는 없다. 다만 배가 움직인다는 물리적 증거(항적)가 부재."
     }
    ]
   },
   "reverse_raw": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 6,
      "verdict_ko": "시선과 조타실 구조는 정확하고 깁스도 명확하지만, 항적이 없어 '멀어지는 중'이라는 이동 증거가 약하고 인물이 걷는 동작이라 '상체를 튼 채 시선 고정'이라는 지시 및 인물이 조타실에 겹쳐 전신 실루엣이 덜 읽힌다."
     },
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "허리 높이 뒤쪽 카메라의 와이드에서 인물 전신이 중앙우측 중경에 놓이고 좌상단 금월도로 상체를 틀어 애틋한 시선을 고정, 흰 항적이 이별의 이동감까지 담아 프레임 레이아웃을 가장 충실히 구현(팔 고정이 캐스트가 아닌 슬링인 점은 감점)."
     }
    ],
    "all_candidates_fail": false,
    "dual": {
     "models": [
      "claude-opus",
      "gpt"
     ],
     "normalized": {
      "A": 1.5,
      "B": 2.0
     },
     "adjusted": {
      "A": 1.5,
      "B": 2.0
     },
     "violations": {},
     "per_model_winner": {
      "claude-opus": "B",
      "gpt": "B"
     },
     "agreed": true
    }
   },
   "reverse_normalized": {
    "winner": "A",
    "ranking": [
     "A",
     "B"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 6,
      "verdict_ko": "시선과 조타실 구조는 정확하고 깁스도 명확하지만, 항적이 없어 '멀어지는 중'이라는 이동 증거가 약하고 인물이 걷는 동작이라 '상체를 튼 채 시선 고정'이라는 지시 및 인물이 조타실에 겹쳐 전신 실루엣이 덜 읽힌다."
     },
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "허리 높이 뒤쪽 카메라의 와이드에서 인물 전신이 중앙우측 중경에 놓이고 좌상단 금월도로 상체를 틀어 애틋한 시선을 고정, 흰 항적이 이별의 이동감까지 담아 프레임 레이아웃을 가장 충실히 구현(팔 고정이 캐스트가 아닌 슬링인 점은 감점)."
     }
    ],
    "all_candidates_fail": false
   },
   "combined": {
    "totals": {
     "A": 16,
     "B": 11
    },
    "ranking": [
     "A",
     "B"
    ],
    "agreement": true,
    "policy": 1
   }
  },
  "readings": [
   {
    "label": "A",
    "direction": "정인우는 프레임 중앙(약간 왼쪽~중앙)에 서서 상체와 시선을 왼쪽 뒤편, 즉 배 항적 너머 멀어지는 섬/부두 방향으로 고정하고 있다. 시선은 화면 좌상단의 섬·방파제 라인에 정확히 착지하며, 카메라는 그의 뒤·측면에서 허리 높이로 잡아 등과 옆얼굴이 보인다. 애틋한 감정이 눈꺼풀과 다문 입에서 읽힌다.",
    "built_space": "어선 갑판: 조타실은 화면 오른쪽, 조타실 유리창 안에 조향 휠 1개만 보이고 계기판·레버가 위치 그대로다. 좌현 난간, 청·백 도장 불워크, 젖은 갑판 등 로케이션 레퍼런스와 일치. 조타실 문·환기 그릴·마스트 등 부속물 개수와 배치가 참조와 맞다. 사람이 앉는 자리는 없고 갑판 위에 서 있어 무리 없다.",
    "entities": "인물은 20대 중반 한국 남성, 검은 단발머리, 참조 얼굴과 상당히 유사. 왼팔은 하늘색 팔걸이 슬링에 걸쳐 고정되어 있으나 '캐스트(깁스)'라기보다 슬링만 보인다. 배경의 섬·방파제·붉은 크레인·건물이 로케이션 사진과 동일. 바다에 항적 거품이 있어 이동 중임이 읽힌다. 텍스트·워터마크 없음.",
    "hard_violations": [],
    "physics": "두 발이 젖은 갑판에 확실히 닿아 체중을 지지하며, 오른팔은 자연스레 내려오고 왼팔은 슬링이 지탱한다. 공중에 뜬 물체 없음. 배의 항적이 진행 방향과 일치해 이동 플랫폼으로 성립. 그림자와 젖은 갑판 반사가 물리적으로 타당."
   },
   {
    "label": "B",
    "direction": "정인우는 프레임 중앙에서 걷는 자세로 몸통은 오른쪽(선수 방향)으로 향하고 고개만 왼쪽 뒤로 돌려 섬 쪽을 본다. 시선은 대체로 화면 좌측 섬 방향으로 향하나 다소 카메라·중경 쪽으로 흐르고, 시선 라인이 상단 좌측의 부두에 정확히 꽂히기보다 애매하다. 카메라는 측면에 가깝고 그의 뒤가 아니라 옆에서 잡아 '뒤에서 따라가는' 지시와 어긋난다.",
    "built_space": "조타실 오른쪽, 조향 휠 1개, 계기판·레버 정상. 좌현 난간과 불워크 배치 일치. 다만 인물이 조타실 문 바로 앞을 가리며 문 손잡이 부분이 어색하게 겹친다. 해수면에 항적 거품이 없어 정박에 가까운 잔잔한 수면 — '이동 중' 증거가 약하다.",
    "entities": "20대 남성, 검은 머리로 참조와 유사하나 얼굴형·이목구비가 레퍼런스와 다소 차이. 오른팔에 흰 붕대/깁스가 감겨 있어 캐스트 조건은 충족. 배경 섬·붉은 크레인·부두는 로케이션과 일치. 텍스트 없음.",
    "hard_violations": [],
    "physics": "양발이 갑판에 닿아 있고 걷는 중간 자세로 무게 이동이 보이나, 팔은 깁스한 채 몸 앞으로 어색하게 접혀 있고 왼손 위치가 불분명하게 조타실 창에 겹친다. 부유하는 물체는 없다. 다만 배가 움직인다는 물리적 증거(항적)가 부재."
   }
  ],
  "totals": {
   "A": 16,
   "B": 11
  },
  "selected": "A",
  "ranking": [
   "A",
   "B"
  ],
  "verdicts": [
   {
    "label": "A",
    "score": 8,
    "verdict_ko": "허리 높이·뒤쪽 카메라, 좌상단 멀어지는 섬으로 고정된 애틋한 시선, 항적으로 드러나는 이동 중 갑판까지 지시를 충실히 구현했다. 다만 인물이 지정된 '중앙 우측'보다 약간 왼쪽이고 깁스가 슬링으로 대체된 점이 감점."
   },
   {
    "label": "B",
    "score": 5,
    "verdict_ko": "깁스는 명확하나 카메라가 뒤가 아닌 측면이고, 잔잔한 수면에 항적이 없어 '멀어지는 중'이라는 이동 증거가 사라졌으며 시선도 섬에 정확히 고정되지 않아 핵심 스테이징에서 밀린다."
   }
  ],
  "refs": [
   {
    "label": "LOCATION PHOTOGRAPH — the exact place of this shot: its architecture, materials, fixed features and lighting mood are spatial truth; stage the moment inside this place. Never copy its camera framing.",
    "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/episodes/d6a9aa85-b75e-400c-980c-4ee7e876a15b/images/background_chain/L15B07.png"
   },
   {
    "label": "CHARACTER REFERENCE — 정인우: the exact person appearing in this shot; match face, hair and build exactly.",
    "path": "<bytes:787232>"
   }
  ],
  "critique": {
   "issues": [
    {
     "issue_ko": "인물의 오른팔을 받치는 슬링과 깁스가 거대한 파란색 일회용 마스크 형태로 잘못 렌더링되었습니다 (심각).",
     "fix_en": "Replace the giant blue face mask on the man's right arm with a realistic white medical plaster cast supported by a standard dark fabric arm sling. Preserve the man's face, his pose, his clothing, the boat deck, the sea, and the background island exactly as they are.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "인물의 의상이 캐릭터 레퍼런스의 녹색 코트와 버건디 터틀넥이 아닌 카키색 재킷과 흰 셔츠로 잘못 그려졌습니다 (심각).",
     "fix_en": "Change the man's khaki jacket and white shirt into a dark green wool overcoat worn over a burgundy turtleneck. Preserve the man's identity, his facial expression, his pose, the cast/sling on his arm, the boat deck, the water, and the background scenery exactly as they are.",
     "needs_regeneration": false,
     "unfixable": false
    },
    {
     "issue_ko": "발이 뱃머리를 향한 상태에서 상체만 튼 스케치의 자세와 달리, 몸 전체가 다소 왼쪽으로 돌아가 있습니다 (주요).",
     "fix_en": "",
     "needs_regeneration": false,
     "unfixable": false
    }
   ]
  },
  "repair_mode": "edit",
  "fix_ref_count": 4,
  "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the giant blue face mask on the man's right arm with a realistic white medical plaster cast supported by a standard dark fabric arm sling. Preserve the man's face, his pose, his clothing, the boat deck, the sea, and the background island exactly as they are.\n- Change the man's khaki jacket and white shirt into a dark green wool overcoat worn over a burgundy turtleneck. Preserve the man's identity, his facial expression, his pose, the cast/sling on his arm, the boat deck, the water, and the background scenery exactly as they are.\n- \n\nPRESERVE EVERYTHING ELSE EXACTLY as in the original photograph —\nevery person, pose, face, garment, object, set feature, lighting,\ncamera and framing stay identical. No text, captions or\nwatermarks anywhere.",
  "fix_rejudge": {
   "forward_raw": {
    "all_candidates_fail": false,
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "readings": [
     {
      "label": "A",
      "direction": "인물은 프레임 중앙우측 갑판에 서서 상체를 좌측(선미/섬 방향)으로 틀고 시선을 화면 좌상단 부두·금월도 쪽으로 고정하고 있다. 시선의 착지점은 좌측 배경의 방파제와 섬으로, 스케치의 좌향 화살표와 일치한다. 다만 머리 각도가 다소 정면에 가까워 '애틋하게 응시'라기보다 멍한 정면 응시에 가깝다.",
      "built_space": "조타실은 배경 레퍼런스대로 1채, 창 안에 조타륜 1개, 계기판과 스로틀 레버가 보인다. 좌현 스테인리스 난간, 흰 불워크 상단, 청색 도장 띠, 우현 뱃머리 난간과 흰 보관함 모두 유지. 다만 카메라가 배경보다 약간 좁게 크롭되어 우현 선수 영역과 갑판 전경이 줄었고, 좌측 부두가 원본보다 크게 잡혔다.",
      "entities": "정인우: 20대 중반 한국 남성, 검은 머리, 캐릭터 레퍼런스와 얼굴 유사. 그러나 의상이 레퍼런스의 짙은 녹색 롱코트 대신 카키/올리브 짧은 재킷으로 바뀌었고 버건디 터틀넥도 안 보인다. 팔걸이는 재킷 안쪽에 옅은 하늘색 천이 살짝 보일 뿐 캐스트/슬링이 거의 식별되지 않아 CARRIED STATE(깁스 고정) 이행이 약하다. 신발은 갈색 구두로 일치.",
      "hard_violations": [],
      "physics": "두 발 모두 젖은 갑판 위에 있고 그림자가 발밑에 붙어 있다. 체중은 뒷발에 실린 자연스러운 스탠스. 부유 요소 없음. 물결·항적은 이동 중 선박에 부합."
     },
     {
      "label": "B",
      "direction": "인물은 중앙우측 갑판, 상체를 좌측으로 비틀고 얼굴과 시선이 화면 좌상단 부두/섬 쪽 수평선에 고정되어 있다. 시선 착지점은 멀어지는 부두와 섬. 스케치 화살표 방향과 일치하고 A보다 고개가 더 확실히 좌측으로 돌아가 '뒤돌아보는 이별의 응시'가 읽힌다.",
      "built_space": "조타실 1채, 창 안 조타륜 1개, 계기판·레버 정상. 좌현 난간·불워크·청색 띠, 우현 선수 난간·흰 보관함 모두 배경 레퍼런스와 거의 동일한 프레이밍과 원근으로 유지. 배경 대비 크롭도 최소.",
      "entities": "정인우: 캐릭터 레퍼런스의 짙은 녹색 울 롱코트 + 버건디 터틀넥 + 회색 슬랙스 + 갈색 구두 그대로. 얼굴·헤어도 레퍼런스와 부합. 오른팔은 흰 캐스트와 회청색 슬링으로 명확히 고정되어 CARRIED STATE 충족.",
      "hard_violations": [],
      "physics": "두 발이 갑판에 접지, 발밑 반사와 그림자 존재. 코트 자락과 옷깃이 자연스럽게 드리우고 슬링이 목에서 지지된다. 부유 요소 없음."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "프레이밍·시선 방향은 대체로 맞으나 캐릭터 레퍼런스의 녹색 롱코트·버건디 터틀넥을 임의로 카키 재킷으로 바꿨고, 필수 지속 항목인 깁스/슬링이 거의 보이지 않으며 크롭도 배경보다 좁아져 손실이 크다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "배경의 카메라·구조물을 그대로 지키면서 레퍼런스 의상(녹색 롱코트·버건디 터틀넥)과 흰 캐스트+슬링을 정확히 재현하고, 상체를 틀어 금월도를 응시하는 스케치 방향까지 충실히 구현했다."
     }
    ]
   },
   "forward_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 5,
      "verdict_ko": "프레이밍·시선 방향은 대체로 맞으나 캐릭터 레퍼런스의 녹색 롱코트·버건디 터틀넥을 임의로 카키 재킷으로 바꿨고, 필수 지속 항목인 깁스/슬링이 거의 보이지 않으며 크롭도 배경보다 좁아져 손실이 크다."
     },
     {
      "label": "B",
      "score": 9,
      "verdict_ko": "배경의 카메라·구조물을 그대로 지키면서 레퍼런스 의상(녹색 롱코트·버건디 터틀넥)과 흰 캐스트+슬링을 정확히 재현하고, 상체를 틀어 금월도를 응시하는 스케치 방향까지 충실히 구현했다."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "인물은 프레임 중앙우측 갑판에 서서 상체를 좌측(선미/섬 방향)으로 틀고 시선을 화면 좌상단 부두·금월도 쪽으로 고정하고 있다. 시선의 착지점은 좌측 배경의 방파제와 섬으로, 스케치의 좌향 화살표와 일치한다. 다만 머리 각도가 다소 정면에 가까워 '애틋하게 응시'라기보다 멍한 정면 응시에 가깝다.",
      "built_space": "조타실은 배경 레퍼런스대로 1채, 창 안에 조타륜 1개, 계기판과 스로틀 레버가 보인다. 좌현 스테인리스 난간, 흰 불워크 상단, 청색 도장 띠, 우현 뱃머리 난간과 흰 보관함 모두 유지. 다만 카메라가 배경보다 약간 좁게 크롭되어 우현 선수 영역과 갑판 전경이 줄었고, 좌측 부두가 원본보다 크게 잡혔다.",
      "entities": "정인우: 20대 중반 한국 남성, 검은 머리, 캐릭터 레퍼런스와 얼굴 유사. 그러나 의상이 레퍼런스의 짙은 녹색 롱코트 대신 카키/올리브 짧은 재킷으로 바뀌었고 버건디 터틀넥도 안 보인다. 팔걸이는 재킷 안쪽에 옅은 하늘색 천이 살짝 보일 뿐 캐스트/슬링이 거의 식별되지 않아 CARRIED STATE(깁스 고정) 이행이 약하다. 신발은 갈색 구두로 일치.",
      "hard_violations": [],
      "physics": "두 발 모두 젖은 갑판 위에 있고 그림자가 발밑에 붙어 있다. 체중은 뒷발에 실린 자연스러운 스탠스. 부유 요소 없음. 물결·항적은 이동 중 선박에 부합."
     },
     {
      "label": "B",
      "direction": "인물은 중앙우측 갑판, 상체를 좌측으로 비틀고 얼굴과 시선이 화면 좌상단 부두/섬 쪽 수평선에 고정되어 있다. 시선 착지점은 멀어지는 부두와 섬. 스케치 화살표 방향과 일치하고 A보다 고개가 더 확실히 좌측으로 돌아가 '뒤돌아보는 이별의 응시'가 읽힌다.",
      "built_space": "조타실 1채, 창 안 조타륜 1개, 계기판·레버 정상. 좌현 난간·불워크·청색 띠, 우현 선수 난간·흰 보관함 모두 배경 레퍼런스와 거의 동일한 프레이밍과 원근으로 유지. 배경 대비 크롭도 최소.",
      "entities": "정인우: 캐릭터 레퍼런스의 짙은 녹색 울 롱코트 + 버건디 터틀넥 + 회색 슬랙스 + 갈색 구두 그대로. 얼굴·헤어도 레퍼런스와 부합. 오른팔은 흰 캐스트와 회청색 슬링으로 명확히 고정되어 CARRIED STATE 충족.",
      "hard_violations": [],
      "physics": "두 발이 갑판에 접지, 발밑 반사와 그림자 존재. 코트 자락과 옷깃이 자연스럽게 드리우고 슬링이 목에서 지지된다. 부유 요소 없음."
     }
    ]
   },
   "reverse_raw": {
    "ranking": [
     "A",
     "B"
    ],
    "winner": "A",
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "A",
      "direction": "정인우는 갑판 좌현 난간 옆에 서서 상체를 왼쪽(카메라 기준 좌측 후방)으로 틀고 시선을 프레임 좌상단의 부두·섬 쪽으로 고정. 시선 목표는 실제로 화면 왼쪽 위의 방파제와 녹색 섬(금월도)에 정확히 착지한다. 카메라는 갑판 뒤쪽 허리 높이에서 그를 뒤/측면으로 잡고 있어 지시와 일치.",
      "built_space": "조타실은 배경 레퍼런스와 동일하게 1동, 문 1개, 창 너머 조타륜 1개(정확한 수), 계기판·레버 확인. 우현 난간, 뱃머리 좌석, 통풍구, 파란 선체 띠 모두 배경과 일치. 좌현 난간 라인과 항적 거품도 유지. 추가·중복 구조물 없음.",
      "entities": "20대 중반 한국인 남성, 어두운 단발, 캐릭터 레퍼런스와 근접한 이목구비. 의상은 레퍼런스의 짙은 녹색 울 코트 + 버건디 터틀넥 + 짙은 회색 슬랙스 + 갈색 구두로 일치. 왼팔은 팔걸이(슬링)+캐스트로 고정되어 CARRIED STATE 충족. 배경의 부두·빨간 등대탑·건물들은 배경 레퍼런스와 동일.",
      "hard_violations": [],
      "physics": "두 발 모두 젖은 갑판에 접지, 오른발 앞·왼발 뒤로 체중이 실린 자연스러운 스탠스. 오른팔은 자연스럽게 늘어뜨리고 왼팔은 슬링이 지지. 코트 자락이 중력대로 드리워짐. 부유하거나 지지 없는 요소 없음."
     },
     {
      "label": "B",
      "direction": "정인우는 갑판 중앙 우측에 서서 상체를 좌측으로 틀고 시선을 좌상단 섬/부두 쪽으로 향함. 시선 목표는 대체로 맞으나 얼굴이 A보다 정면에 가깝고 시선이 다소 위쪽 허공으로 흐른다. 카메라 위치는 유사.",
      "built_space": "조타실 1동, 조타륜 1개, 우현 난간·좌석 등 배경 요소는 유지. 다만 좌현 난간 기둥/핸드레일이 인물 뒤에서 잘려 흐려지고, 좌현 난간 상단 레일이 인물 앞에서 연결이 어색하게 끊긴 부분이 있음. 부두 뒤 등대탑·건물은 유지되나 배경 대비·해상도가 전반적으로 뭉개져 있음.",
      "entities": "20대 중반 한국인 남성으로 얼굴은 레퍼런스와 유사. 그러나 의상이 레퍼런스의 짙은 녹색 롱코트가 아니라 짧은 카키/올리브 사파리 재킷과 흰 이너로 교체됨 — 계약이 요구하지 않은 변경. 팔걸이는 연한 하늘색 천으로 재킷 안쪽에 어정쩡하게 걸려 있어 캐스트 여부가 불분명.",
      "hard_violations": [],
      "physics": "두 발 모두 갑판에 접지되어 지지는 정상. 다만 왼팔이 슬링 천 뒤로 사실상 사라져 팔의 부피가 보이지 않고, 오른팔은 몸통에 붙어 어색. 재킷 하단과 슬링의 연결이 물리적으로 모호하며 옷 주름의 무게감이 A보다 약함."
     }
    ],
    "verdicts": [
     {
      "label": "A",
      "score": 8,
      "verdict_ko": "레퍼런스의 녹색 롱코트·버건디 터틀넥 의상과 얼굴을 정확히 재현하고, 좌상단 금월도로 상체를 튼 애틋한 시선·슬링 고정 팔·배경 조타실 및 난간 기하까지 모두 일치하는 충실한 스테이징."
     },
     {
      "label": "B",
      "score": 4,
      "verdict_ko": "시선 방향과 접지는 무난하나 계약이 요구하지 않은 의상 교체(녹색 롱코트→카키 짧은 재킷)로 인물 정체성을 훼손했고, 팔걸이·캐스트 표현이 모호하며 좌현 난간과 배경 디테일도 뭉개져 손실이 이득을 넘어섬."
     }
    ]
   },
   "reverse_normalized": {
    "winner": "B",
    "ranking": [
     "B",
     "A"
    ],
    "verdicts": [
     {
      "label": "B",
      "score": 8,
      "verdict_ko": "레퍼런스의 녹색 롱코트·버건디 터틀넥 의상과 얼굴을 정확히 재현하고, 좌상단 금월도로 상체를 튼 애틋한 시선·슬링 고정 팔·배경 조타실 및 난간 기하까지 모두 일치하는 충실한 스테이징."
     },
     {
      "label": "A",
      "score": 4,
      "verdict_ko": "시선 방향과 접지는 무난하나 계약이 요구하지 않은 의상 교체(녹색 롱코트→카키 짧은 재킷)로 인물 정체성을 훼손했고, 팔걸이·캐스트 표현이 모호하며 좌현 난간과 배경 디테일도 뭉개져 손실이 이득을 넘어섬."
     }
    ],
    "all_candidates_fail": false,
    "readings": [
     {
      "label": "B",
      "direction": "정인우는 갑판 좌현 난간 옆에 서서 상체를 왼쪽(카메라 기준 좌측 후방)으로 틀고 시선을 프레임 좌상단의 부두·섬 쪽으로 고정. 시선 목표는 실제로 화면 왼쪽 위의 방파제와 녹색 섬(금월도)에 정확히 착지한다. 카메라는 갑판 뒤쪽 허리 높이에서 그를 뒤/측면으로 잡고 있어 지시와 일치.",
      "built_space": "조타실은 배경 레퍼런스와 동일하게 1동, 문 1개, 창 너머 조타륜 1개(정확한 수), 계기판·레버 확인. 우현 난간, 뱃머리 좌석, 통풍구, 파란 선체 띠 모두 배경과 일치. 좌현 난간 라인과 항적 거품도 유지. 추가·중복 구조물 없음.",
      "entities": "20대 중반 한국인 남성, 어두운 단발, 캐릭터 레퍼런스와 근접한 이목구비. 의상은 레퍼런스의 짙은 녹색 울 코트 + 버건디 터틀넥 + 짙은 회색 슬랙스 + 갈색 구두로 일치. 왼팔은 팔걸이(슬링)+캐스트로 고정되어 CARRIED STATE 충족. 배경의 부두·빨간 등대탑·건물들은 배경 레퍼런스와 동일.",
      "hard_violations": [],
      "physics": "두 발 모두 젖은 갑판에 접지, 오른발 앞·왼발 뒤로 체중이 실린 자연스러운 스탠스. 오른팔은 자연스럽게 늘어뜨리고 왼팔은 슬링이 지지. 코트 자락이 중력대로 드리워짐. 부유하거나 지지 없는 요소 없음."
     },
     {
      "label": "A",
      "direction": "정인우는 갑판 중앙 우측에 서서 상체를 좌측으로 틀고 시선을 좌상단 섬/부두 쪽으로 향함. 시선 목표는 대체로 맞으나 얼굴이 A보다 정면에 가깝고 시선이 다소 위쪽 허공으로 흐른다. 카메라 위치는 유사.",
      "built_space": "조타실 1동, 조타륜 1개, 우현 난간·좌석 등 배경 요소는 유지. 다만 좌현 난간 기둥/핸드레일이 인물 뒤에서 잘려 흐려지고, 좌현 난간 상단 레일이 인물 앞에서 연결이 어색하게 끊긴 부분이 있음. 부두 뒤 등대탑·건물은 유지되나 배경 대비·해상도가 전반적으로 뭉개져 있음.",
      "entities": "20대 중반 한국인 남성으로 얼굴은 레퍼런스와 유사. 그러나 의상이 레퍼런스의 짙은 녹색 롱코트가 아니라 짧은 카키/올리브 사파리 재킷과 흰 이너로 교체됨 — 계약이 요구하지 않은 변경. 팔걸이는 연한 하늘색 천으로 재킷 안쪽에 어정쩡하게 걸려 있어 캐스트 여부가 불분명.",
      "hard_violations": [],
      "physics": "두 발 모두 갑판에 접지되어 지지는 정상. 다만 왼팔이 슬링 천 뒤로 사실상 사라져 팔의 부피가 보이지 않고, 오른팔은 몸통에 붙어 어색. 재킷 하단과 슬링의 연결이 물리적으로 모호하며 옷 주름의 무게감이 A보다 약함."
     }
    ]
   },
   "combined": {
    "totals": {
     "A": 9,
     "B": 17
    },
    "ranking": [
     "B",
     "A"
    ],
    "agreement": true,
    "policy": 1
   },
   "winner": "B",
   "fix_won": true,
   "all_candidates_fail": false,
   "policy": "fix_rejudge_v2_orig_on_tie"
  },
  "bgfirst": {
   "bg_path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/e716bafb-24bb-42b7-aea0-fdb383844ee8/images/d6a9aa85-b75e-400c-980c-4ee7e876a15b/scene/recipe/S113sh2__bgfirst_bg.png",
   "bg_asset_id": "0d368f14-ed64-4cfd-b5dc-5935e7a983ac",
   "bg_record_key": "S113sh2::bgfirst_bg",
   "chain_winner": true,
   "authority": "plate"
  },
  "ref_mode": "재투영 배경+콘티+엔티티 (2택1: 체인 승)",
  "share_plan": {
   "ref_plan": "background"
  }
 }
}