{ "S1sh1": { "input_fingerprint": "bfc455a3aa258fb4", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 한쪽 신발이 벗겨진 한국인 남자가 우거진 수풀 사이를 달리며 한 발을 허공에 내딛고 있는 도중의 전신 구도\n\nLOCATION (lock): A dense forest with overgrown underbrush forming a rough escape 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\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The fleeing man is missing one shoe, and the large bite wound on his thigh is visibly bleeding.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 한국인 남자 (Korean 성인 남성, 짧은 검은 머리, 평범한 얼굴형, 일반적인 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": 5 }, "selected": "A", "ranking": [ "A", "C", "B" ], "verdicts": [ { "label": "A", "score": 7, "verdict_ko": "황혼의 시간대와 허공에 발을 내딛는 역동적인 동작, 허벅지의 상처 및 벗겨진 신발 등 프롬프트의 요구사항을 가장 정확하게 구현했습니다." }, { "label": "B", "score": 4, "verdict_ko": "상처와 맨발 상태는 잘 표현되었으나, 조명이 황혼이라기보다 다소 차갑고 어두우며 동작의 역동성이 떨어집니다." }, { "label": "C", "score": 5, "verdict_ko": "황혼의 조명과 소품 상태는 적절하지만, 앞으로 내딛는 발이 땅에 닿아 있어 '허공에 내딛는 도중'이라는 지시와 맞지 않습니다." } ], "refs": [ { "label": "CHARACTER REFERENCE — 한국인 남자: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "허공에 내디딘 맨발(오른발)의 형태가 기형적이며, 발가락과 발뒤꿈치의 구조가 해부학적으로 어긋나 있습니다.", "fix_en": "Redraw the raised bare right foot to have correct human anatomy and natural orientation." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the raised bare right foot to have correct human anatomy and natural orientation.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S1sh4": { "input_fingerprint": "dc8ee917f2c15359", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 검은 형상의 인영이 허공에 뛰어오른 채 한국인 남자를 향해 두 팔을 뻗고 있는 순간\n\nLOCATION (lock): A dense forest with overgrown underbrush forming a rough escape path through the vegetation. 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 dusk forest, tangled undergrowth, and dim natural light from the reference. Exclude the fleeing man's exact running pose and detached shoe unless they fall naturally within this attack 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 man remains one-shoed with the bleeding bite wound on his thigh as the dark figure lunges at him.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 한국인 남자 (Korean 성인 남성, 짧은 검은 머리, 평범한 얼굴형, 일반적인 이목구비); 검은 인영 (인종과 성별을 식별할 수 없는 인간형, 보이지 않는 얼굴 세부, 검은 머리 윤곽, 실루엣 형태의 머리와 얼굴) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": 4, "C": 7 }, "selected": "C", "ranking": [ "C", "A", "B" ], "verdicts": [ { "label": "A", "score": 5, "verdict_ko": "남자의 오른발에 신발이 신겨져 있어 이전 샷의 상태(오른쪽 맨발)를 유지하라는 지시를 어겼습니다." }, { "label": "B", "score": 4, "verdict_ko": "검은 인영이 전술 마스크와 장비를 착용하여 실루엣 묘사 지시와 어긋나며, 남자의 신발 상태도 틀렸습니다." }, { "label": "C", "score": 7, "verdict_ko": "남자의 한쪽 맨발(오른발)과 허벅지 상처가 이전 샷과 정확히 이어지며, 덮쳐오는 검은 인영의 행동과 형태도 지시에 부합합니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/scene/recipe/S1sh1_sel.png" }, { "label": "CHARACTER REFERENCE — 한국인 남자: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "남자의 맨발과 등산화를 신은 발의 좌우 위치가 레퍼런스(오른발 맨발, 왼발 등산화)와 반대로 바뀌어 있습니다.", "fix_en": "Swap the man's footwear so that his right foot is bare and his left foot is wearing the hiking boot." }, { "issue_ko": "검은 인영의 머리와 얼굴이 단순한 실루엣이 아니라 눈구멍과 주름이 보이는 복면(발라클라바)을 쓴 형태로 묘사되었습니다.", "fix_en": "Render the dark figure's head and face as a solid, featureless black silhouette without any mask holes or fabric details." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Swap the man's footwear so that his right foot is bare and his left foot is wearing the hiking boot.\n- Render the dark figure's head and face as a solid, featureless black silhouette without any mask holes or fabric details.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+prev+엔티티" }, "S2sh1": { "input_fingerprint": "37d1df0ac98de10e", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night, dim streetlights.\n\nSHOT TEXT (authoritative, Korean): 희미한 가로등 불빛 아래 전봇대에 밀착된 채 서 있는 한국인 남자와 한국인 여자의 옆모습\n\nLOCATION (lock): A secluded urban back alley with utility poles and dim streetlights. 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 man and woman remain pressed together against the utility pole beneath the dim streetlight.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 한국인 여자 (Korean 성인 여성, 자연스러운 검은 머리, 평범한 타원형 얼굴, 일반적인 이목구비); 한국인 남자 (포식 상태) (Korean 성인 남성, 짧은 검은 머리, 평범한 얼굴형, 자연스러운 기본 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": 6, "B": 4, "C": 3 }, "selected": "A", "ranking": [ "A", "B", "C" ], "verdicts": [ { "label": "A", "score": 6, "verdict_ko": "남자의 비니와 코트 등 캐릭터 지침을 유일하게 잘 준수했으나, 전봇대에 완전히 밀착된 포즈가 아니며 금지된 텍스트(상자 로고)가 존재합니다." }, { "label": "B", "score": 4, "verdict_ko": "남자의 비니가 누락되었고 전봇대가 두 사람 사이를 가로막고 있어 포즈 지침을 어겼으며, 전봇대에 금지된 텍스트가 뚜렷하게 나타납니다." }, { "label": "C", "score": 3, "verdict_ko": "두 사람이 전봇대에 밀착된 포즈는 가장 잘 표현했으나, 남자의 비니가 없고 의상(짧은 가죽 재킷)이 참조 이미지와 완전히 달라 캐릭터 일관성이 떨어집니다." } ], "refs": [ { "label": "CHARACTER REFERENCE — 한국인 여자: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "CHARACTER REFERENCE — 한국인 남자 (포식 상태): the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "남자와 여자가 서로 밀착되어 있지 않고 눈에 띄게 떨어져 서 있습니다.", "fix_en": "Close the gap between the man and the woman so they are physically pressed tightly together against the utility pole." }, { "issue_ko": "화면 왼쪽 아래의 파란색 상자에 'Cass'라는 영어 텍스트가 명확하게 보입니다.", "fix_en": "Remove the 'Cass' text from the blue crates on the bottom left corner." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Close the gap between the man and the woman so they are physically pressed tightly together against the utility pole.\n- Remove the 'Cass' text from the blue crates on 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": "플레이트+콘티+엔티티" }, "S2sh3": { "input_fingerprint": "52b332c415245b77", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night, dim streetlights.\n\nSHOT TEXT (authoritative, Korean): 한국인 남자가 입을 크게 벌려 한국인 여자의 목에 이빨을 박아넣은 채 멈춘 순간\n\nLOCATION (lock): A secluded urban back alley with utility poles and dim streetlights. 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 streetlamp lighting, utility pole, and shadowy alley textures from the reference. Exclude the earlier face-to-face posture and show only the positions required for the neck bite.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\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 pair remain against the utility pole, with the man's teeth embedded in the woman's neck and his eyes red.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 한국인 여자 (Korean 성인 여성, 자연스러운 검은 머리, 평범한 타원형 얼굴, 일반적인 이목구비); 한국인 남자 (포식 상태) (Korean 성인 남성, 짧은 검은 머리, 평범한 얼굴형, 자연스러운 기본 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": 2 }, "selected": "A", "ranking": [ "A", "B", "C" ], "verdicts": [ { "label": "A", "score": 7, "verdict_ko": "이전 샷의 위치(오른쪽 전봇대에 기댄 남자)를 정확히 유지하며, 입을 크게 벌려 목을 무는 행동과 붉은 눈을 지시대로 잘 표현했습니다." }, { "label": "B", "score": 3, "verdict_ko": "남녀의 위치가 반대로 바뀌어 이전 샷의 설정을 위반했으며, 목을 무는 핵심 행동과 붉은 눈이 누락되었습니다." }, { "label": "C", "score": 2, "verdict_ko": "전봇대 위치가 왼쪽으로 바뀌었고 여자가 기대어 있어 연속성이 깨졌으며, 무는 행동도 전혀 나타나지 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/scene/recipe/S2sh1_sel.png" }, { "label": "CHARACTER REFERENCE — 한국인 여자: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "CHARACTER REFERENCE — 한국인 남자 (포식 상태): the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "남자가 여자의 목을 물고 있어야 하지만, 이미지에서는 반대로 여자의 얼굴이 남자의 뺨에 밀착되어 있고 남자의 입은 허공을 향해 벌려져 있습니다.", "fix_en": "Change the pose so the man's face is positioned against the woman's neck, with his mouth wide open and his teeth embedded in her skin." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the pose so the man's face is positioned against the woman's neck, with his mouth wide open and his teeth embedded in her 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.", "ref_mode": "플레이트+prev+엔티티" }, "S3sh3": { "input_fingerprint": "a6ab277115544778", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 형사 무리 한가운데 서서 손가락으로 한쪽을 가리키는 혜수의 상체\n\nLOCATION (lock): A daytime crime-scene montage spanning a forest trail and an urban back alley, both secured with police lines as officers examine the ground and remove bodies. 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: 혜수 (Korean 40대 중반 여성, 단정한 검은 머리, 성숙한 타원형 얼굴, 또렷한 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": 5, "C": 4 }, "selected": "A", "ranking": [ "A", "B", "C" ], "verdicts": [ { "label": "A", "score": 7, "verdict_ko": "숲과 골목이 혼재된 배경 및 형사 무리 중앙이라는 구도를 가장 잘 구현했으나, 금지된 텍스트가 포함되어 감점되었습니다." }, { "label": "B", "score": 5, "verdict_ko": "형사 무리 중앙에 있다는 느낌이 약하며 장소의 디테일이 부족하고 금지된 텍스트가 나타납니다." }, { "label": "C", "score": 4, "verdict_ko": "도심 골목 배경이 누락되었으며, 의상과 소품에 금지된 문자와 숫자가 명확히 보입니다." } ], "refs": [ { "label": "CHARACTER REFERENCE — 혜수: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "폴리스 라인, 뒤편의 밴 차량, 형사들의 신분증에 텍스트가 포함되어 있음.", "fix_en": "Remove all text from the police tape, vehicles, and ID badges." } ] }, "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 police tape, vehicles, and ID badges.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S4sh1": { "input_fingerprint": "2f71f0a3358ffdd3", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 맑은 낮, 오래된 다세대 빌라 옥상 위에 자리 잡은 옥탑방을 높은 곳에서 내려다본 부감 구도\n\nLOCATION (lock): The rooftop level of an old low-rise villa on the outskirts of a coastal city, centered on a small rooftop dwelling. 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": { "C": 7, "A": 5, "B": 4 }, "selected": "C", "ranking": [ "C", "A", "B" ], "verdicts": [ { "label": "C", "score": 7, "verdict_ko": "가스 계량기 위치 등 레퍼런스의 구조를 가장 정확히 재현하며 해안 도시의 부감 구도를 잘 살렸습니다." }, { "label": "A", "score": 5, "verdict_ko": "부감 구도는 우수하나 옥탑방 정면에 원본과 다른 잘못된 구조물이 생성되었습니다." }, { "label": "B", "score": 4, "verdict_ko": "옥탑방 외벽에 지시사항에 위배되는 텍스트 형태의 아티팩트가 포함되었습니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L03B02.png" } ], "critique": { "issues": [ { "issue_ko": "레퍼런스의 카메라 구도를 절대 복사하지 말라는 지시를 어기고 동일한 프레이밍을 사용했습니다.", "fix_en": "Change the camera framing to a completely different high-angle perspective." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the camera framing to a completely different high-angle perspective.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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)" }, "S5sh5": { "input_fingerprint": "3ec570d71da49a30", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day, thin sunlight through a small window.\n\nSHOT TEXT (authoritative, Korean): 다급하게 상체를 기울이며 손가락 끝으로 TV 전원 버튼을 세게 누르고 있는 민숙의 측면 구도\n\nLOCATION (lock): A compact rooftop apartment interior with a small living-room window, a kitchen sink, a stove with a boiling pot, a dining table, and adjoining bedrooms. 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): Min-suk already bears the small circular mark on her wrist, though it need not be exposed in this angle.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 민숙 (Korean 40대 초반 여성, 자연스러운 검은 머리, 성숙한 타원형 얼굴, 평범하고 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, "A": 4, "C": 4 }, "selected": "B", "ranking": [ "B", "A", "C" ], "verdicts": [ { "label": "B", "score": 7, "verdict_ko": "캐릭터 레퍼런스(헤어스타일, 의상)와 정확히 일치하며 TV 하단의 전원 버튼을 누르는 핵심 동작을 완벽하게 구현했습니다." }, { "label": "A", "score": 4, "verdict_ko": "캐릭터의 헤어스타일과 의상이 레퍼런스와 다르며, 전원 버튼이 아닌 TV 상단 모서리를 누르고 있어 동작 묘사가 부정확합니다." }, { "label": "C", "score": 4, "verdict_ko": "캐릭터의 외모가 레퍼런스와 일치하지 않고, TV 전원이 아닌 셋톱박스 쪽으로 손을 뻗고 있어 지시문과 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B04.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S5sh5.png" }, { "label": "CHARACTER REFERENCE — 민숙: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "캐릭터 레퍼런스 이미지에 명시된 베이지색 앞치마를 착용하고 있지 않습니다.", "fix_en": "Add the beige apron around the character's waist to match the reference." }, { "issue_ko": "TV 화면에 반사된 모습이 상체를 기울인 인물의 실제 자세 및 의상(분홍색 티셔츠)과 전혀 다른 똑바로 선 형태입니다.", "fix_en": "Update the reflection on the TV screen to correctly show the character's leaning posture and pink shirt." }, { "issue_ko": "오른쪽 손목이 화면에 노출되어 있음에도 설정에 명시된 작은 원형 자국이 보이지 않습니다.", "fix_en": "Add the small circular mark to her exposed right wrist." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the beige apron around the character's waist to match the reference.\n- Update the reflection on the TV screen to correctly show the character's leaning posture and pink shirt.\n- Add the small circular mark to her exposed right wrist.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S5sh9": { "input_fingerprint": "b232fd8ea6ac81ef", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day, thin sunlight through a small window.\n\nSHOT TEXT (authoritative, Korean): 민숙의 손목에 선명하게 드러난 작은 원형 표식의 타이트 샷\n\nLOCATION (lock): A compact rooftop apartment interior with a small living-room window, a kitchen sink, a stove with a boiling pot, a dining table, and adjoining bedrooms. 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 small circular mark remains clearly visible on Min-suk's wrist.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 민숙 (Korean 40대 초반 여성, 자연스러운 검은 머리, 성숙한 타원형 얼굴, 평범하고 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": 6, "A": 4 }, "selected": "B", "ranking": [ "B", "C", "A" ], "verdicts": [ { "label": "B", "score": 7, "verdict_ko": "손목의 선명한 원형 표식과 더불어 배경에 지정된 끓는 냄비와 주방 구조가 참조와 완벽하게 일치합니다." }, { "label": "C", "score": 6, "verdict_ko": "타이트 샷의 구도와 아파트 배경은 적절하나, 주방의 세부 묘사(끓는 냄비 등)가 부족합니다." }, { "label": "A", "score": 4, "verdict_ko": "지정된 의상 색상은 보이지만, 손목을 쥐고 있는 손가락의 해부학적 구조가 크게 왜곡되었습니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B04.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S5sh9.png" }, { "label": "CHARACTER REFERENCE — 민숙: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "화면에 보이는 손의 피부가 심하게 주름지고 거칠게 묘사되어, 40대 초반 여성인 민숙의 신체적 특징과 일치하지 않습니다.", "fix_en": "Render the hands with smoother, more feminine skin that matches an early-40s woman." }, { "issue_ko": "우측 상단에 위치한 손의 손가락들이 서로 뭉개져 있고 관절이 기형적으로 꺾여 있습니다.", "fix_en": "Correct the fingers on the top-right hand to have distinct, anatomically normal structures and joints." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Render the hands with smoother, more feminine skin that matches an early-40s woman.\n- Correct the fingers on the top-right hand to have distinct, anatomically normal structures and joints.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S6sh1": { "input_fingerprint": "3fc5a729bbe740db", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 낮의 해안가 도로 위, 이어폰을 낀 채 자전거 안장에 앉아 한 발로 페달을 밟아 내리는 중간 동작으로 미소 짓는 수리영의 옆모습, 자전거의 추진력에 의해 머리카락이 뒤로 흩날리는 찰나\n\nLOCATION (lock): A coastal road running beside the sea, suitable for bicycle travel. 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): Su-ri-young is wearing her earphones while riding the bicycle.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": 6, "A": 5 }, "selected": "C", "ranking": [ "C", "B", "A" ], "verdicts": [ { "label": "C", "score": 8, "verdict_ko": "한 발로 페달을 밟아 내리는 중간 동작이 프레임 내에 명확히 보이며, 흩날리는 머리카락과 옆모습 등 프롬프트의 요구사항을 가장 잘 구현했습니다." }, { "label": "B", "score": 6, "verdict_ko": "캐릭터와 배경의 표현은 좋으나, 페달을 밟는 하반신 동작이 프레임 밖으로 잘려 명시된 행동을 온전히 보여주지 못합니다." }, { "label": "A", "score": 5, "verdict_ko": "페달을 밟는 발이 화면에 포함되지 않았으며, 핸들을 쥔 손과 자전거의 구조가 다소 부자연스럽게 묘사되었습니다." } ], "refs": [ { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "자전거 반대편에 있는 왼쪽 발이 페달 위에 있지 않고 체인링 아래 허공에 비정상적으로 매달려 있습니다.", "fix_en": "Place the character's background left foot securely onto the left bicycle pedal." }, { "issue_ko": "레퍼런스 이미지에서 시계는 왼쪽 손목에 착용되어 있으나, 생성된 이미지에서는 오른쪽 손목에 착용되어 있습니다.", "fix_en": "Remove the watch from the character's right wrist and place it on her left wrist." }, { "issue_ko": "핸들을 쥐고 있는 오른쪽 손의 손가락 형태가 기형적으로 왜곡되어 있습니다.", "fix_en": "Correct the anatomy of the right hand so it has a normal five-finger grip securely holding the handlebar." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Place the character's background left foot securely onto the left bicycle pedal.\n- Remove the watch from the character's right wrist and place it on her left wrist.\n- Correct the anatomy of the right hand so it has a normal five-finger grip securely holding the handlebar.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S7sh3": { "input_fingerprint": "f7094850f03aee05", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 로트와일러가 귀를 뒤로 젖힌 채 얌전히 바닥에 엎드린 정지 상태\n\nLOCATION (lock): A stray-dog shelter with secured enclosures for multiple dogs and working areas for vaccinations, grooming, ointment treatment, and bathing. 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 Rottweiler remains subdued and lying calmly after responding to Su-ri-young's stare.\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": "텍스트 배제 지시를 준수했으며, 귀를 젖히고 얌전히 엎드린 로트와일러의 모습을 지시에 맞게 잘 구현했습니다." }, { "label": "B", "score": 6, "verdict_ko": "텍스트 금지 지시는 따랐으나, 로트와일러의 귀가 자연스럽게 앞으로 쳐져 있어 '귀를 뒤로 젖힌' 디테일이 부족합니다." }, { "label": "C", "score": 4, "verdict_ko": "로트와일러의 자세는 잘 표현되었으나, 배경에 텍스트가 명확히 포함되어 '텍스트 금지'라는 중요한 배제 지시를 위반했습니다." } ], "refs": [], "critique": { "issues": [ { "issue_ko": "개의 귀가 뒤로 젖혀지지 않고 양옆으로 처져 있습니다.", "fix_en": "Flatten and pin the dog's ears tightly back against its head." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Flatten and pin the dog's ears tightly back against its 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": "플레이트만 (배경 전용)" }, "S7sh6": { "input_fingerprint": "6d661a605724c67c", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 수리영이 귀에 스마트폰을 댄 채 미간을 찌푸린 얼굴\n\nLOCATION (lock): A stray-dog shelter with secured enclosures for multiple dogs and working areas for vaccinations, grooming, ointment treatment, and bathing. 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 shelter's kennel materials, indoor lighting, and subdued utilitarian look from the reference. Exclude the rottweiler's prominent foreground placement and the injection moment from this phone-call 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): Su-ri-young has her smartphone with her and is using it to call Min-suk.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": "인물, 표정, 배경을 정확히 재현했으며 후보 중 유일하게 스마트폰의 형태가 정상적입니다." }, { "label": "A", "score": 5, "verdict_ko": "인물과 배경 묘사는 우수하나, 스마트폰 뒷면에 하단 포트가 생성되는 구조적 오류가 있습니다." }, { "label": "C", "score": 4, "verdict_ko": "전반적인 지시사항은 잘 따랐으나, 스마트폰 뒷면에 충전 단자와 스피커가 잘못 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/scene/recipe/S7sh3_sel.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "PROP REFERENCE — 스마트폰: the exact object appearing in this shot; match its look, material and wear exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "폰을 쥔 왼손의 손가락 방향이 뒤집혀 해부학적으로 어긋나며, 스마트폰의 화면(기능적 면)이 귀가 아닌 카메라 쪽을 향해 거꾸로 들려 있습니다.", "fix_en": "Redraw the left hand with correct anatomy so the fingers wrap from the outer edge inwards, and flip the smartphone so its screen faces her ear and its back faces the camera." }, { "issue_ko": "캐릭터 레퍼런스 이미지에서 수리영이 착용하고 있는 회색 모자가 누락되었습니다.", "fix_en": "Add the grey cap from the character reference onto her head." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw the left hand with correct anatomy so the fingers wrap from the outer edge inwards, and flip the smartphone so its screen faces her ear and its back faces the camera.\n- Add the grey cap from the character reference onto her 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": "플레이트+prev+엔티티" }, "S8sh3": { "input_fingerprint": "81a4bbb1babeb04c", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk, neon signs.\n\nSHOT TEXT (authoritative, Korean): 수리영이 마트 유니폼을 입은 한국인 여성 직원을 마주보며 입을 살짝 벌린 측면 구도\n\nLOCATION (lock): A busy commercial district beside a harbor, filled with neon signs, taxis, street vendors, and food stalls. It connects to a large supermarket with merchandise aisles and an exterior bicycle-parking 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\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Su-ri-young's bicycle remains parked outside the mart, and she still has her smartphone with 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: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": 6, "C": 5 }, "selected": "A", "ranking": [ "A", "B", "C" ], "verdicts": [ { "label": "A", "score": 7, "verdict_ko": "로케이션 사진의 실내외 구조와 외부 벽돌 건물 등 고정 요소들을 가장 정확히 재현했으나, 지문에 없는 인물들이 매장 내부에 등장해 일부 감점되었습니다." }, { "label": "B", "score": 6, "verdict_ko": "인물의 외양과 소품은 지문과 일치하나, 창밖의 지정된 건축물이 전혀 다른 상가 건물로 변형되어 공간적 일치도가 다소 떨어집니다." }, { "label": "C", "score": 5, "verdict_ko": "행동 지문은 잘 구현되었지만 창밖으로 항구가 바로 보이고 매장 내부 진열장 구조가 달라져 로케이션 사진의 공간적 사실성을 크게 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L08B01.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S8sh3.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "샷 텍스트에 명시되지 않은 인물들(마트 내부 배경의 남성 2명 및 창밖 거리의 사람들)이 등장함.", "fix_en": "Remove all background people from inside the supermarket and the street outside." }, { "issue_ko": "직원 유니폼의 명찰과 소매, 마트 내부의 종이 상자, 창밖 간판 등에 텍스트가 포함되어 있음.", "fix_en": "Remove all text and writing from the employee's uniform, the cardboard boxes, and the exterior signs." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all background people from inside the supermarket and the street outside.\n- Remove all text and writing from the employee's uniform, the cardboard boxes, and the exterior signs.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S8sh5": { "input_fingerprint": "7b29a8b47fb5bce9", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk, neon signs.\n\nSHOT TEXT (authoritative, Korean): 마트 밖으로 나온 수리영이 손에 든 스마트폰 액정을 내려다보는 상체\n\nLOCATION (lock): A busy commercial district beside a harbor, filled with neon signs, taxis, street vendors, and food stalls. It connects to a large supermarket with merchandise aisles and an exterior bicycle-parking 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\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The bicycle remains outside for Su-ri-young's return, and she is holding the smartphone used to call Min-suk.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": "지정된 장소의 건축물과 캐릭터의 의상을 가장 완벽하게 재현했으나, 지문에 없는 추가 인물 및 텍스트 금지 규칙을 위반했습니다." }, { "label": "A", "score": 5, "verdict_ko": "배경 구조는 참조와 유사하나, 캐릭터가 지정되지 않은 데님 재킷을 입고 있으며 지문 외 인물과 텍스트가 포함되었습니다." }, { "label": "B", "score": 4, "verdict_ko": "캐릭터 의상은 일치하지만, 필수 조건인 장소 레퍼런스(특정 모퉁이 건축물 및 구조)를 완전히 이탈했습니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L08B02.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S8sh5.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "PROP REFERENCE — 스마트폰: the exact object appearing in this shot; match its look, material and wear exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "샷 텍스트에 언급되지 않은 다른 사람들이 배경(길거리, 포장마차 등)에 포함되어 있습니다.", "fix_en": "Remove all background people so only Su-ri-young is visible in the shot." }, { "issue_ko": "수리영이 들고 있는 스마트폰의 뒷면(카메라 렌즈 부분)이 수리영의 시선을 향하고 있습니다.", "fix_en": "Flip the smartphone so its screen faces Su-ri-young's eyes." }, { "issue_ko": "네온사인과 간판에 한글 텍스트가 생성되었습니다.", "fix_en": "Remove all text and lettering from the neon signs and storefronts." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all background people so only Su-ri-young is visible in the shot.\n- Flip the smartphone so its screen faces Su-ri-young's eyes.\n- Remove all text and lettering from the neon signs and storefronts.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S9sh1": { "input_fingerprint": "069c9de334932101", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 어두운 해안도로 위 자전거 안장에 앉아 한쪽 다리로 페달을 밟아 내리는 중간 자세로 앞을 향해 상체를 기울인 수리영의 실루엣, 옷자락이 뒤로 휘날리는 순간\n\nLOCATION (lock): A coastal road following the dark shoreline, used as a bicycle route toward nearby homes. 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): Su-ri-young continues toward home on the same bicycle, with her smartphone still on her person.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": "안장에 앉아 페달을 밟는 자세, 흩날리는 옷자락, 뒷주머니의 스마트폰까지 모든 지시사항을 완벽하게 충족했습니다." }, { "label": "C", "score": 5, "verdict_ko": "지정된 자세와 배경은 준수했으나, 요구된 스마트폰이 인물에게서 보이지 않습니다." }, { "label": "A", "score": 4, "verdict_ko": "안장에 앉지 않고 서서 페달을 밟고 있으며, 스마트폰이 누락되었고 참조 이미지에 없는 옷을 허리에 둘렀습니다." } ], "refs": [ { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "프롬프트는 '자전거 안장에 앉아'라고 명시했으나, 이미지에서 인물은 엉덩이를 안장에서 떼고 공중에 띄운 채 일어서 있습니다.", "fix_en": "Seat the character firmly with her buttocks resting entirely on the bicycle saddle." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Seat the character firmly with her buttocks resting entirely on the bicycle saddle.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S10sh6": { "input_fingerprint": "66b27e1f85e6d91e", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 현관 철문이 아주 살짝 열린 틈새로 보이는 어두운 실내와 신발 앞코\n\nLOCATION (lock): A rooftop terrace around a small rooftop apartment, with a water tank, clothesline, drying platform, scattered dried herbs, and a metal entrance door. 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 stranger has forced the lock with the leaflet, leaving the iron entrance door slightly open as he enters.\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": 4, "A": 3 }, "selected": "C", "ranking": [ "C", "B", "A" ], "verdicts": [ { "label": "C", "score": 7, "verdict_ko": "밤 시간대와 옥상 배경을 잘 구현했으나, 신발이 문 틈새 안쪽이 아닌 바깥에 놓인 점이 아쉽습니다." }, { "label": "B", "score": 4, "verdict_ko": "야간 시간대 조건을 어겼으며, 인물이 등장하지 않아야 한다는 지시와 달리 사람의 다리가 보입니다." }, { "label": "A", "score": 3, "verdict_ko": "카메라 시점이 실내에서 밖을 바라보는 방향으로 완전히 뒤바뀌었고, 문이 너무 넓게 열려 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L03B03.png" } ], "critique": { "issues": [ { "issue_ko": "'열린 틈새로 보이는 신발 앞코'라는 지시와 달리, 신발 전체가 문 안쪽이 아닌 야외 바닥에 여러 개 놓여 있습니다.", "fix_en": "Remove the shoes lying on the terrace floor and show only the tip of a shoe visible inside the dark room through the vertical crack of the slightly open door." }, { "issue_ko": "텍스트 생성을 금지했으나 문고리에 꽂힌 전단지에 가짜 글자들이 빼곡히 적혀 있습니다.", "fix_en": "Remove all text and pseudo-text characters from the leaflet attached to the door." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the shoes lying on the terrace floor and show only the tip of a shoe visible inside the dark room through the vertical crack of the slightly open door.\n- Remove all text and pseudo-text characters from the leaflet attached to 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": "플레이트만 (배경 전용)" }, "S12sh11": { "input_fingerprint": "ffb297d2fc5bffd7", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night, dim floor lamp and television glow.\n\nSHOT TEXT (authoritative, Korean): 옷과 피부가 피투성이가 된 민숙이 바닥에 주저앉아 고개를 푹 숙인 정지 상태\n\nLOCATION (lock): A compact rooftop apartment with a living room, dining table, small kitchen, main bedroom, and a curtained second bedroom. The rooms contain a cooled cooking pot, a television, a floor lamp, and simple domestic furniture. 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): Minsuk is seated motionless behind the curtain, with her head dropped forward and one hand tightly clutching a crumpled photograph. The source does not specify her facing direction, torso angle, leg positions, other arm, or supporting contact.\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): Min-suk's corpse remains seated motionless behind the curtain, with her head dropped forward and one hand tightly clutching a crumpled photograph, her shoulder and collarbone torn open, and the small wrist mark present. Because she cannot move, the crumpled photograph tightly clutched in one hand must already be there; blood pools, spatters, and bloody footprints remain across the room.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 민숙 (환시) (Korean 40대 초반 여성, 자연스러운 검은 머리, 성숙한 타원형 얼굴, 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": "캐릭터 레퍼런스(의상, 외모)와 피투성이 상태, 중력에 순응하는 축 늘어진 자세를 가장 정확하게 구현했습니다." }, { "label": "A", "score": 3, "verdict_ko": "캐릭터의 외모는 비슷하나, 프롬프트에 명시된 피투성이 상태, 찢어진 의상, 바닥의 핏자국이 완전히 누락되었습니다." }, { "label": "B", "score": 3, "verdict_ko": "유혈 묘사는 존재하지만, 지정된 캐릭터(단발, 붉은 셔츠)가 아닌 레이아웃 스케치의 인물(긴 머리, 원피스)을 잘못 묘사했습니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B03.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S12sh11.png" }, { "label": "CHARACTER REFERENCE — 민숙 (환시): the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "민숙이 커튼 뒤에 있지 않고 방 입구 벽에 기대어 앉아 있습니다.", "fix_en": "Relocate Minsuk so she is seated motionless behind the curtain." }, { "issue_ko": "시간대 설정에 명시된 TV 불빛(television glow)이 없이 화면이 완전히 꺼져 있습니다.", "fix_en": "Add a dim glow emitting from the television screen." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Relocate Minsuk so she is seated motionless behind the curtain.\n- Add a dim glow emitting from the television screen.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S12sh18": { "input_fingerprint": "fdf80f59ff1f6214", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night, dim floor lamp and television glow.\n\nSHOT TEXT (authoritative, Korean): 민숙의 손에 꽉 쥐어진 구겨진 사진의 타이트 샷\n\nLOCATION (lock): A compact rooftop apartment with a living room, dining table, small kitchen, main bedroom, and a curtained second bedroom. The rooms contain a cooled cooking pot, a television, a floor lamp, and simple domestic furniture. 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 dim bedroom lighting, bloodied condition, and the corpse's fixed seated position from the reference. Exclude the full body and surrounding room from the tight frame, retaining only the clenched hand and crumpled photograph.\n\nIMMOBILE CHARACTER POSE — CANONICAL (identical wherever this character appears in ANY panel; on any conflict THIS POSE WINS): Minsuk is seated motionless behind the curtain, with her head dropped forward and one hand tightly clutching a crumpled photograph. The source does not specify her facing direction, torso angle, leg positions, other arm, or supporting contact.\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): Min-suk remains dead, seated motionless behind the curtain, with her head dropped forward and one hand tightly clutching a crumpled photograph, amid the unchanged bloodstains, with the circular mark on her wrist.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 민숙 (환시) (Korean 40대 초반 여성, 자연스러운 검은 머리, 성숙한 타원형 얼굴, 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": 6 }, "selected": "A", "ranking": [ "A", "C", "B" ], "verdicts": [ { "label": "A", "score": 8, "verdict_ko": "꽉 쥐어 구겨진 사진의 상태와 무릎에 얹힌 중력 묘사, TV 불빛까지 모든 프롬프트 요구사항을 훌륭하게 충족했습니다." }, { "label": "B", "score": 4, "verdict_ko": "TV 불빛 등 배경은 맞으나, '꽉 쥐어진 구겨진 사진'이라는 핵심 행동 지침을 무시하고 사진을 느슨하게 들고 있습니다." }, { "label": "C", "score": 6, "verdict_ko": "사진을 꽉 쥔 손의 묘사와 자세는 훌륭하지만, 지시된 TV 불빛(television glow)이 꺼져 있어 조명 조건에 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B03.png" }, { "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/scene/recipe/S12sh11_sel.png" }, { "label": "CHARACTER REFERENCE — 민숙 (환시): the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "PROP REFERENCE — 구겨진 피 묻은 낡은 흑백 사진: the exact object appearing in this shot; match its look, material and wear exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "프롬프트에서 '주변 공간을 프레임에서 제외하라(Exclude the surrounding room)'고 명시했으나, 배경에 거실의 TV와 조명이 그대로 보입니다.", "fix_en": "Remove the background living room elements (TV, lamp) to make it a completely tight frame containing only the hand, photograph, and immediate lap area." }, { "issue_ko": "사진의 앞면이 카메라를 향해 정면으로 보여지고 있어, 죽은 인물이 자신의 눈으로 보던 방향이 아니라 시청자를 위해 의도적으로 제시하는(presenting) 것처럼 연출되었습니다.", "fix_en": "Rotate the photograph so its front surface faces inward towards the character's body or upward, rather than presenting it directly to the camera." }, { "issue_ko": "엄지손가락의 살과 구겨진 사진의 윗부분 모서리가 서로 녹아내려 경계선 없이 하나의 물체처럼 융합되어 있습니다.", "fix_en": "Render a clear, distinct boundary between the skin of the thumb and the paper edge of the photograph so they do not melt together." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the background living room elements (TV, lamp) to make it a completely tight frame containing only the hand, photograph, and immediate lap area.\n- Rotate the photograph so its front surface faces inward towards the character's body or upward, rather than presenting it directly to the camera.\n- Render a clear, distinct boundary between the skin of the thumb and the paper edge of the photograph so they do not melt together.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+prev+엔티티" }, "S13sh4": { "input_fingerprint": "058fafc3c597fcbe", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 수리영이 초점을 잃은 눈으로 허공을 응시하며 서 있는 모습\n\nLOCATION (lock): The exterior of a rooftop apartment atop an old multi-unit villa, reached by a steep outdoor stairway from the yard below. 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): Su-ri-young still has her smartphone with her after fleeing the apartment, while her bicycle remains in the building's rear yard.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": "지시된 자세와 초점 잃은 표정, 소지품 상태를 정확히 구현했으며, 위치 레퍼런스의 임시 객체(경찰차)를 적절히 제외하여 공간을 사실적으로 묘사했습니다." }, { "label": "B", "score": 5, "verdict_ko": "전반적인 배경과 캐릭터의 묘사는 우수하나, 스마트폰을 쥐고 있는 왼손의 형태가 다소 왜곡되어 아쉽습니다." }, { "label": "A", "score": 4, "verdict_ko": "캐릭터와 소지품의 상태는 양호하나, 건축물과 고정 요소만 반영해야 하는 위치 지침을 어기고 레퍼런스의 경찰차를 그대로 포함했습니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L10B01.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S13sh4.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "왼쪽 담벼락에 있는 자전거의 앞바퀴가 콘크리트 벽 구조물을 뚫고 들어가 있습니다.", "fix_en": "Render the bicycle's front wheel so it rests naturally against the surface of the wall without clipping through the solid concrete." }, { "issue_ko": "캐릭터의 왼손(화면 오른쪽) 손가락들이 제대로 분리되지 않고 뭉개진 덩어리처럼 표현되어 있습니다.", "fix_en": "Redraw the character's left hand to show clearly defined, anatomically correct fingers instead of a merged shape." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Render the bicycle's front wheel so it rests naturally against the surface of the wall without clipping through the solid concrete.\n- Redraw the character's left hand to show clearly defined, anatomically correct fingers instead of a merged 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": "플레이트+콘티+엔티티" }, "S10sh7": { "input_fingerprint": "247de2b173f13902", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 불이 켜진 옥탑방 창문에 거대한 그림자가 작은 그림자의 어깨 쪽으로 고개를 꺾은 형태가 맺혀 있는 정지 구도\n\nLOCATION (lock): A rooftop terrace around a small rooftop apartment, with a water tank, clothesline, drying platform, scattered dried herbs, and a metal entrance door. 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 forced entrance door remains open, and Min-suk still bears the small circular wrist mark as the stranger attacks her shoulder.\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": "창문에 맺힌 거대한 그림자가 작은 그림자의 어깨 쪽으로 고개를 꺾은 형태를 인물 노출 없이 정확히 구현했습니다." }, { "label": "A", "score": 6, "verdict_ko": "그림자 형태로 지시를 따랐으나, 그림자 간의 거대한 크기 차이와 고개를 꺾는 디테일이 다소 부족합니다." }, { "label": "B", "score": 3, "verdict_ko": "창문에 맺힌 그림자만 보여야 한다는 지시와 달리 인물의 실제 모습이 유리창 너머로 명확히 보여 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L03B03.png" } ], "critique": { "issues": [] }, "fix_skipped": true, "ref_mode": "플레이트만 (배경 전용)" }, "S11sh4": { "input_fingerprint": "134b1861e8b055a5", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 철제 파이프 난간을 손으로 잡은 수리영이 코끝을 찡그린 채 계단 중간에 서 있는 얼굴 클로즈업\n\nLOCATION (lock): A narrow residential alley leading into the yard of an old multi-unit villa, with a rear bicycle area and an exterior stairway climbing to a rooftop apartment. 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): Su-ri-young's bicycle has been left in the building's rear yard, while the rooftop entrance door remains half open above 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: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": 2 }, "selected": "B", "ranking": [ "B", "A", "C" ], "verdicts": [ { "label": "B", "score": 7, "verdict_ko": "지시된 코 찡그린 표정과 한 손으로 난간을 잡은 포즈를 오류 없이 가장 자연스럽게 구현했습니다." }, { "label": "A", "score": 5, "verdict_ko": "요구된 표정은 반영되었으나, 레퍼런스 및 스케치와 달리 두 손으로 난간을 잡은 점이 아쉽습니다." }, { "label": "C", "score": 2, "verdict_ko": "인물의 왼쪽 어깨와 몸통이 배경의 벽돌 벽과 겹쳐 투명하게 렌더링된 심각한 구조적 오류가 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L11B01.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S11sh4.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "건물 뒷마당(배경의 오른쪽 계단 뒤편)에 명시된 수리영의 자전거가 없습니다.", "fix_en": "Add Su-ri-young's bicycle parked in the rear yard behind the stairs." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add Su-ri-young's bicycle parked in the rear yard behind the stairs.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S11sh5": { "input_fingerprint": "327ef05bf529f06a", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 옥상에 선 수리영의 시선 끝에 반쯤 열려 있는 어두운 현관 철문이 보이는 정면 구도\n\nLOCATION (lock): A narrow residential alley leading into the yard of an old multi-unit villa, with a rear bicycle area and an exterior stairway climbing to a rooftop apartment. 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 bicycle remains roughly parked in the rear yard, and the rooftop iron entrance door is still half open.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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 위치, 인물의 뒷모습 정면 구도, 반쯤 열린 문과 배경의 자전거까지 모든 요소를 충실히 구현했습니다." }, { "label": "B", "score": 5, "verdict_ko": "인물과 배경의 구도는 일치하나, 프롬프트에 명시된 배경의 자전거가 누락되었습니다." }, { "label": "C", "score": 4, "verdict_ko": "자전거와 인물은 존재하지만 레퍼런스에 없는 조명이 문 위에 임의로 추가되어 장소의 일관성을 해칩니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/map_plate_S11sh5.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S11sh5.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "프롬프트 텍스트에 자전거는 뒷마당(rear yard)에 주차되어 있다고 명시되어 있으나 옥상에 잘못 배치되어 있습니다.", "fix_en": "Remove the bicycle from the rooftop." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the bicycle from the rooftop.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S12sh21": { "input_fingerprint": "8ee52b3fb097a4dc", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night, dim floor lamp and television glow.\n\nSHOT TEXT (authoritative, Korean): 침대 위쪽 벽면에 붉은색 원형 표식이 절반쯤 나타난 상태\n\nLOCATION (lock): A compact rooftop apartment with a living room, dining table, small kitchen, main bedroom, and a curtained second bedroom. The rooms contain a cooled cooking pot, a television, a floor lamp, and simple domestic furniture. 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): Min-suk's corpse and the room's blood evidence remain in place, but the photograph has now been removed from her hand and her wrist mark has turned red. A red circle is forming on the wall above the bed.\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": 4, "B": 3 }, "selected": "C", "ranking": [ "C", "A", "B" ], "verdicts": [ { "label": "C", "score": 7, "verdict_ko": "지정된 레퍼런스의 공간 구조와 핏자국을 정확하게 유지하면서 절반만 나타난 붉은 원형 표식을 성공적으로 구현했습니다." }, { "label": "A", "score": 4, "verdict_ko": "벽면의 붉은 표식이 절반을 넘어 거의 완성된 상태이며, 레퍼런스의 공간 구조와 벽면 배치가 크게 왜곡되었습니다." }, { "label": "B", "score": 3, "verdict_ko": "레퍼런스의 실내 건축 구조를 전혀 반영하지 못했고, 붉은 표식 또한 지시와 다르게 완전히 온전한 원형으로 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B03.png" } ], "critique": { "issues": [ { "issue_ko": "TV 화면이 꺼져 있어 프롬프트에 명시된 TV 불빛(television glow)이 전혀 없습니다.", "fix_en": "Turn on the television screen so it emits a visible glow into the room." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Turn on the television screen so it emits a visible glow into the room.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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)" }, "S13sh6": { "input_fingerprint": "f42e6b3c6d1ab533", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 고개를 든 채 혜수의 얼굴을 마주보고 있는 수리영의 측면 구도\n\nLOCATION (lock): The exterior of a rooftop apartment atop an old multi-unit villa, reached by a steep outdoor stairway from the yard below. 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 dim courtyard, building exterior, and emergency-lit night atmosphere from the reference. Exclude the earlier vacant stare and any vehicles or officers outside this tighter two-person 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): Su-ri-young remains outside with her smartphone, and her bicycle is still parked in the rear yard as Hye-su prepares to go upstairs.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태); 혜수 (Korean 40대 중반 여성, 단정한 검은 머리, 성숙한 타원형 얼굴, 또렷한 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": 5, "C": 2 }, "selected": "A", "ranking": [ "A", "B", "C" ], "verdicts": [ { "label": "A", "score": 7, "verdict_ko": "두 인물의 외양과 복장이 레퍼런스와 정확히 일치하며, 지시된 측면 구도와 디테일이 충실하게 구현되었습니다." }, { "label": "B", "score": 5, "verdict_ko": "구도와 배경은 적절하나, 혜수의 비니가 누락되고 헤어스타일이 레퍼런스와 다르게 묘사되었습니다." }, { "label": "C", "score": 2, "verdict_ko": "'얼굴을 마주보고'라는 텍스트를 직역하여 혜수를 공중에 뜬 머리로만 묘사함으로써 실사 인간 형태 지침을 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L10B01.png" }, { "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/scene/recipe/S13sh4_sel.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "CHARACTER REFERENCE — 혜수: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "이전 샷의 넓은 카메라 구도를 그대로 복사하지 말고 타이트한 2인 구도로 변경하여 차량을 배제하라는 지시를 어기고, 와이드 프레이밍을 유지하여 배경의 자동차와 오토바이가 여전히 노출되었습니다.", "fix_en": "Push the camera in to create a tight two-shot of the characters, completely framing out the gate and the vehicles in the background." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Push the camera in to create a tight two-shot of the characters, completely framing out the gate and the vehicles 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": "플레이트+prev+엔티티" }, "S14sh5": { "input_fingerprint": "aeec9fcf52b61252", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 핏자국 없이 깨끗한 방 내부 전경\n\nLOCATION (lock): A compact rooftop apartment whose living area, dining table, and bedroom now appear unnaturally clean and orderly. A backpack and a doll sit in the bedroom in subtly altered positions. 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 bedroom has been inexplicably reset: Min-suk's corpse, every bloodstain and footprint, and the red wall circle are gone, leaving the room excessively clean and orderly.\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": 4, "B": 3 }, "selected": "C", "ranking": [ "C", "A", "B" ], "verdicts": [ { "label": "C", "score": 7, "verdict_ko": "레퍼런스의 패널 벽과 타일 바닥, 창밖 풍경을 잘 유지하면서 혈흔과 시체 흔적을 지우고 가방과 인형을 정확히 배치했습니다." }, { "label": "A", "score": 4, "verdict_ko": "벽과 바닥 재질이 완전히 달라졌으며, 지시와 달리 커튼 아래에 시체 형태가 여전히 남아있습니다." }, { "label": "B", "score": 3, "verdict_ko": "창밖 풍경이 고층 빌딩으로 바뀌어 장소 일관성이 크게 훼손되었고 벽의 질감도 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B03.png" } ], "critique": { "issues": [] }, "fix_skipped": true, "ref_mode": "플레이트만 (배경 전용)" }, "S14sh7": { "input_fingerprint": "ab5deb538545c588", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 수리영이 멍한 눈으로 텅 빈 방 안을 응시하는 얼굴 클로즈업\n\nLOCATION (lock): A compact rooftop apartment whose living area, dining table, and bedroom now appear unnaturally clean and orderly. A backpack and a doll sit in the bedroom in subtly altered positions. 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 spotless bedroom, unchanged room geometry, and subdued night lighting from the reference. Exclude any corpse, blood, red wall mark, or attack debris, and keep the background secondary to the 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): The room remains spotless and empty, with no corpse, blood evidence, or red wall mark anywhere.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": "캐릭터의 의상(초록색 맨투맨)과 표정을 정확히 구현하고 소품(침대 위 인형)의 미세한 변화 및 장소의 세부 요소를 완벽히 유지했으나, 피사계 심도가 깊어 배경이 덜 강조되지 않은 점이 아쉽습니다." }, { "label": "C", "score": 5, "verdict_ko": "지정된 캐릭터와 의상을 잘 따랐으나, 배경의 주요 구조(식탁의 의자 누락, 파티션 창문 왜곡 등)가 심하게 변형되어 장소 일관성(Location lock)을 위반했습니다." }, { "label": "A", "score": 4, "verdict_ko": "얼굴 클로즈업과 얕은 피사계 심도로 카메라 연출은 우수하나, 이전 숏에서 유지되어야 할 캐릭터의 의상(초록색 맨투맨)을 완전히 누락하여 엄격한 복장 제한을 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B03.png" }, { "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/scene/recipe/S14sh5_sel.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "이전 샷 스틸컷에 있던 침실 안쪽의 큰 커튼이 사라지고 창문과 벽이 그대로 노출되었습니다.", "fix_en": "Restore the large curtain covering the left wall and window of the bedroom exactly as it appears in the previous shot still." }, { "issue_ko": "이전 샷 스틸컷에서 바닥에 놓여 있던 인형과 백팩이 침대 위와 침대 옆으로 위치가 바뀌었습니다.", "fix_en": "Move the doll and backpack back to the floor in front of the curtain, matching their exact positions from the previous shot still." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore the large curtain covering the left wall and window of the bedroom exactly as it appears in the previous shot still.\n- Move the doll and backpack back to the floor in front of the curtain, matching their exact positions 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": "플레이트+prev+엔티티" }, "S14sh11": { "input_fingerprint": "43b3886cf0eed228", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 혜수가 날카로운 눈빛으로 김형사를 노려보는 모습\n\nLOCATION (lock): A compact rooftop apartment whose living area, dining table, and bedroom now appear unnaturally clean and orderly. A backpack and a doll sit in the bedroom in subtly altered positions. 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 apartment remains unnaturally clean: the corpse and blood are absent, and the two teacups previously on the table are gone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 혜수 (Korean 40대 중반 여성, 단정한 검은 머리, 성숙한 타원형 얼굴, 또렷한 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": "지시문에 따라 혜수와 김형사를 올바르게 배치하고, 의상 및 유지 상태(혈흔 및 찻잔 제거)를 충실히 반영했습니다." }, { "label": "A", "score": 4, "verdict_ko": "텍스트에 명시된 김형사가 등장하지 않으며, 혜수의 의상(가죽 재킷)과 자세가 레퍼런스와 일치하지 않습니다." }, { "label": "C", "score": 3, "verdict_ko": "전경에 위치한 김형사 인물이 두 명으로 중복 생성되어 인체 묘사 규칙을 심각하게 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B03.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S14sh11.png" }, { "label": "CHARACTER REFERENCE — 혜수: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "백팩이 침실 안이 아니라 거실(침실 문밖) 바닥에 놓여 있습니다.", "fix_en": "Move the backpack from the living room floor so that it sits inside the bedroom." }, { "issue_ko": "바닥 전체에 어두운 얼룩과 핏자국이 남아 있어 '비정상적으로 깨끗하다'는 지시 및 혈흔이 없어야 한다는 설정과 어긋납니다.", "fix_en": "Remove all dark splatters, stains, and grime from the entire floor to make it appear unnaturally clean." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Move the backpack from the living room floor so that it sits inside the bedroom.\n- Remove all dark splatters, stains, and grime from the entire floor to make it appear unnaturally clean.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S15sh2": { "input_fingerprint": "d39cf8986f4ccd1a", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 혜수의 손이 음료수 캔을 쥔 채 수리영 앞으로 내밀어진 순간\n\nLOCATION (lock): A bus-stop area beside the coast, with seating oriented toward the open 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): Su-ri-young still carries her smartphone as Hye-su extends the drink can to 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: 혜수 (Korean 40대 중반 여성, 단정한 검은 머리, 성숙한 타원형 얼굴, 또렷한 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": 4, "A": 3 }, "selected": "B", "ranking": [ "B", "C", "A" ], "verdicts": [ { "label": "B", "score": 7, "verdict_ko": "혜수의 외모와 복장이 참조와 완벽히 일치하며, 캔을 건네는 동작과 상대방의 스마트폰 소지 상태를 가장 정확히 묘사했습니다." }, { "label": "C", "score": 4, "verdict_ko": "캔을 건네는 인물의 의상이 참조 이미지(검은 가죽 재킷)와 일치하지 않아 캐릭터 설정이 어긋났습니다." }, { "label": "A", "score": 3, "verdict_ko": "허공에 떠 있는 절단된 팔이 묘사되어 해부학적으로 심각한 오류가 발생했습니다." } ], "refs": [ { "label": "CHARACTER REFERENCE — 혜수: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "음료수 캔 표면에 프롬프트에서 엄격히 금지한 텍스트(브랜드명 등)가 생성되어 있음.", "fix_en": "Remove all text, letters, and fake branding from the surface of the drink can." }, { "issue_ko": "레퍼런스 이미지에서 혜수가 착용하고 있는 회색 비니가 누락됨.", "fix_en": "Add the dark grey beanie to Hye-su's head to match the reference character exactly." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all text, letters, and fake branding from the surface of the drink can.\n- Add the dark grey beanie to Hye-su's head to match the reference character 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.", "ref_mode": "플레이트+콘티+엔티티" }, "S15sh6": { "input_fingerprint": "0555119e589c6179", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night.\n\nSHOT TEXT (authoritative, Korean): 멀리서 혜수의 전신과 한 발을 공중에 띄운 mid-stride 자세로 달리는 수리영의 뒷모습을 한 프레임에 담은 넓은 구도\n\nLOCATION (lock): A bus-stop area beside the coast, with seating oriented toward the open 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): Su-ri-young runs off with her smartphone still on her person.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태); 혜수 (Korean 40대 중반 여성, 단정한 검은 머리, 성숙한 타원형 얼굴, 또렷한 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": 3 }, "selected": "A", "ranking": [ "A", "B", "C" ], "verdicts": [ { "label": "A", "score": 8, "verdict_ko": "캐릭터의 의상(수리영의 모자, 혜수의 비니), 공중에 발이 뜬 달리는 자세, 스마트폰 소지 등 프롬프트와 레퍼런스를 가장 정확하게 구현했습니다." }, { "label": "B", "score": 5, "verdict_ko": "지정된 달리는 자세는 잘 표현되었으나, 수리영의 모자와 혜수의 비니 등 핵심 의상 디테일이 누락되었습니다." }, { "label": "C", "score": 3, "verdict_ko": "두 캐릭터의 모자/비니가 모두 누락되었고, 역동적인 달리는 자세가 부족하며 스마트폰이 중복 생성되는 오류가 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/map_plate_S15sh6.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S15sh6.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "CHARACTER REFERENCE — 혜수: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "벤치 옆 바닥에 그림자를 드리우는 사람이 없음에도 불구하고 출처를 알 수 없는 사람 형태의 그림자가 잘못 생성되었습니다.", "fix_en": "Remove the detached, human-shaped shadow from the ground next to the bench." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the detached, human-shaped shadow from the ground next to the bench.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S16sh3": { "input_fingerprint": "02426f4798423c98", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): night, office lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 혜수가 스마트폰 액정을 김형사 쪽으로 불쑥 내민 순간\n\nLOCATION (lock): A police-station office with duty desks, computers, telephones, and work areas for investigators. 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): Hye-su is holding her smartphone with Min-suk's saved number displayed on its screen.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 혜수 (Korean 40대 중반 여성, 단정한 검은 머리, 성숙한 타원형 얼굴, 또렷한 이목구비); 김형사 (Korean 30대 남성, 짧은 검은 머리, 성숙한 얼굴, 완만하게 각진 턱선) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": "혜수의 가죽 재킷과 단정한 머리, 스마트폰 소품이 참조와 일치하며 지정된 장소의 구조물을 왜곡 없이 정확히 구현했습니다." }, { "label": "B", "score": 5, "verdict_ko": "캐릭터 의상은 훌륭하나, 원본 장소에 없는 스탠드 조명이 발명되었고 지문에 없는 김형사의 손동작이 추가되었습니다." }, { "label": "A", "score": 4, "verdict_ko": "소품과 행동은 적절히 표현되었으나, 혜수가 가죽 재킷을 입지 않아 캐릭터 참조 이미지와의 일치도가 크게 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L13B02.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S16sh3.png" }, { "label": "CHARACTER REFERENCE — 혜수: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "PROP REFERENCE — 스마트폰: the exact object appearing in this shot; match its look, material and wear exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "스마트폰 화면이 카메라(김형사 쪽)를 향하고 있으나, 혜수의 오른손은 손등이 카메라를 향하고 있어 폰을 쥐는 방향과 화면의 방향이 물리적으로 모순됩니다.", "fix_en": "Redraw Hye-su's right hand with the palm facing the camera so she is correctly holding the smartphone with its screen pointing forward." }, { "issue_ko": "스마트폰의 외형이 프롭 레퍼런스에 제시된 붉은색과 금색 테두리 디자인과 일치하지 않고 단순한 검은색 폰으로 나타납니다.", "fix_en": "Apply the exact red and gold metallic casing from the prop reference to the smartphone." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Redraw Hye-su's right hand with the palm facing the camera so she is correctly holding the smartphone with its screen pointing forward.\n- Apply the exact red and gold metallic casing from the prop reference to the smartphone.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S17sh2": { "input_fingerprint": "6db9db6d03adba2f", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dawn, rooftop apartment light on.\n\nSHOT TEXT (authoritative, Korean): 옥상 담벼락 위에서 한쪽 앞발을 허공으로 뻗은 검은 고양이의 측면\n\nLOCATION (lock): The rooftop exterior of an old villa, with a small rooftop apartment visible beyond a wall; the apartment light remains on. 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 rooftop room remains lit while the black cat crosses the wall.\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": 3 }, "selected": "A", "ranking": [ "A", "B", "C" ], "verdicts": [ { "label": "A", "score": 8, "verdict_ko": "레퍼런스의 장소와 조명을 완벽하게 재현했으며, 앞발을 뻗은 검은 고양이의 측면 모습도 매우 자연스럽습니다." }, { "label": "B", "score": 5, "verdict_ko": "지정된 장소와 시간대는 일치하지만, 허공으로 뻗은 고양이의 앞발이 해부학적으로 심하게 왜곡되었습니다." }, { "label": "C", "score": 3, "verdict_ko": "고양이의 동작은 프롬프트에 부합하나, 필수적으로 유지해야 할 레퍼런스의 건축 구조와 장소를 완전히 다르게 변형했습니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L03B01.png" } ], "critique": { "issues": [ { "issue_ko": "기준 사진에서 조명이 켜진 창문 아래에 위치한 에어컨 실외기와 나무 평상이 누락되었습니다.", "fix_en": "Restore the white air conditioner outdoor unit and the wooden platform below the lit window to match the reference photograph's fixed features." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore the white air conditioner outdoor unit and the wooden platform below the lit window to match the reference photograph's fixed 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.", "ref_mode": "플레이트만 (배경 전용)" }, "S18sh11": { "input_fingerprint": "6a505d488ad26953", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): early morning, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 거울 속 반사된 욕실 타일 벽면에 붉은색 원형 표식이 절반쯤 나타난 상태\n\nLOCATION (lock): A small rooftop apartment containing a main bedroom, a meticulously ordered second bedroom with curtains, a bathroom with a wall mirror, and a living area with a table and framed family photograph. 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 previously reset bedroom remains excessively clean and corpse-free. Su-ri-young has taken out her doll-adorned backpack, and a red circle is temporarily forming on the reflected bathroom wall.\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": 4, "A": 3 }, "selected": "B", "ranking": [ "B", "C", "A" ], "verdicts": [ { "label": "B", "score": 7, "verdict_ko": "거울에 반사된 욕실 타일 벽면에 붉은 원이 '절반쯤' 나타난 상태를 가장 정확히 연출했으나 가방이 프레임에서 누락됨." }, { "label": "C", "score": 4, "verdict_ko": "미완성된 원과 인형 가방을 포함했으나, 원이 욕실 타일이 아닌 거실 벽면에 나타나 지시문을 위반함." }, { "label": "A", "score": 3, "verdict_ko": "가방은 묘사되었으나 붉은 원이 '절반쯤'이 아닌 온전한 형태로 나타나 핵심 상황 지시를 어김." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B01.png" } ], "critique": { "issues": [ { "issue_ko": "레퍼런스의 화장실에는 없는 창문이 왼쪽 벽과 거울 반사면에 나타남", "fix_en": "Remove the window from the left bathroom wall and its reflection in the mirror, replacing them with plain tiles." }, { "issue_ko": "Carried state에 명시된 '인형이 달린 배낭'이 화면에 보이지 않음", "fix_en": "Place a doll-adorned backpack somewhere visible on the bathroom floor or sink counter." }, { "issue_ko": "거울 표면에 프롬프트에 없는 불필요한 테이프 조각이 붙어 있음", "fix_en": "Remove the piece of tape attached to the glass surface of the mirror." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the window from the left bathroom wall and its reflection in the mirror, replacing them with plain tiles.\n- Place a doll-adorned backpack somewhere visible on the bathroom floor or sink counter.\n- Remove the piece of tape attached to the glass surface of the mirror.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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)" }, "S18sh15": { "input_fingerprint": "9ea4de10e67d454c", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): early morning, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 액자 속 민숙의 얼굴 부분 이미지가 아래로 일그러지며 흘러내리는 찰나\n\nLOCATION (lock): A small rooftop apartment containing a main bedroom, a meticulously ordered second bedroom with curtains, a bathroom with a wall mirror, and a living area with a table and framed family photograph. 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): Su-ri-young's doll-adorned backpack and smartphone remain with her as Min-suk's face in the framed photograph temporarily distorts and runs downward.\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 공간을 정확히 구현하면서 주석(빨간 원)을 올바르게 배제했고, 액자 속 얼굴이 흘러내리는 핵심 묘사를 완벽히 표현했습니다." }, { "label": "B", "score": 4, "verdict_ko": "공간과 액자 묘사는 준수하나, 레퍼런스 이미지에 그려진 주석(빨간 원)을 그대로 복사하여 주석 배제 지시를 위반했습니다." }, { "label": "C", "score": 3, "verdict_ko": "액자 속 묘사는 표현되었으나, 레퍼런스 이미지와 전혀 다른 실내 구조와 배경을 생성하여 장소 고정 조건을 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B01.png" } ], "critique": { "issues": [ { "issue_ko": "사진 속 여성의 입과 혀가 비정상적으로 길게 늘어나 해부학적 변형 금지 지침을 위반했으며, 왜곡된 이미지가 평면이 아닌 3D 액체처럼 액자 밖으로 흘러나오고 있습니다.", "fix_en": "Restore the woman's normal facial anatomy in the photograph, and render the downward distortion strictly as 2D printed ink melting on the flat surface inside the frame." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore the woman's normal facial anatomy in the photograph, and render the downward distortion strictly as 2D printed ink melting on the flat surface inside 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": "플레이트만 (배경 전용)" }, "S18sh16": { "input_fingerprint": "22d887c6050bc898", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): early morning, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 수리영이 두 손으로 머리를 감싸 쥐고 입을 크게 벌린 얼굴\n\nLOCATION (lock): A small rooftop apartment containing a main bedroom, a meticulously ordered second bedroom with curtains, a bathroom with a wall mirror, and a living area with a table and framed family photograph. 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 modest living-room look, daylight, and interior materials from the reference. Exclude the earlier television action, apron-clad woman, cookware, and any bathroom elements from this distressed 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): The framed photograph has returned to normal, and the bathroom's red circle has vanished; Su-ri-young still has her doll-adorned backpack and smartphone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": "요구된 인형이 달린 배낭이 포함되었고 조리 도구가 성공적으로 배제되었으나, 액자 속 사진이 여전히 왜곡되어 있습니다." }, { "label": "A", "score": 5, "verdict_ko": "지시된 표정과 의상은 잘 표현되었으나, 제외해야 할 조리 도구가 보이며 요구된 배낭이 존재하지 않습니다." }, { "label": "C", "score": 4, "verdict_ko": "표정은 지시문과 일치하나 인물의 의상이 다르고, 명시된 배낭과 스마트폰이 보이지 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B01.png" }, { "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/scene/recipe/S5sh5_sel.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "테이블 위 액자 속 사진이 정상으로 돌아와야 한다는 지시와 달리 여전히 왜곡된 소용돌이 형태입니다.", "fix_en": "Replace the distorted swirly image in the picture frame on the table with a normal family photograph." }, { "issue_ko": "수리영이 인형이 달린 배낭과 스마트폰을 소지(착용)하고 있어야 하나, 스마트폰은 없고 배낭은 배경의 의자에 놓여 있습니다.", "fix_en": "Render Su-ri-young wearing the doll-adorned backpack and holding her smartphone, and remove the backpack from the chair behind her." }, { "issue_ko": "캐릭터 레퍼런스에 포함된 회색 야구 모자를 착용하고 있지 않습니다.", "fix_en": "Add the dark grey baseball cap from the character reference onto Su-ri-young's head." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the distorted swirly image in the picture frame on the table with a normal family photograph.\n- Render Su-ri-young wearing the doll-adorned backpack and holding her smartphone, and remove the backpack from the chair behind her.\n- Add the dark grey baseball cap from the character reference onto Su-ri-young'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": "플레이트+prev+엔티티" }, "S19sh5": { "input_fingerprint": "5dc93430eb6bf309", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): early morning, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 수리영과 남자직원이 마트 밖에서 벽면의 CCTV를 올려다보는 뒷모습\n\nLOCATION (lock): A large supermarket in the quiet early hours, with broad merchandise aisles, an exterior bicycle area, and outdoor walls covered by surveillance cameras. 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): Su-ri-young's bicycle remains parked outside the mart, and she continues to carry her doll-adorned backpack and smartphone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태); 남자직원 (Korean 중년 남성, 짧고 단정한 검은 머리, 평범한 얼굴형, 중년의 자연스러운 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": 5 }, "selected": "A", "ranking": [ "A", "C", "B" ], "verdicts": [ { "label": "A", "score": 7, "verdict_ko": "요구된 스마트폰과 인형 달린 백팩을 잘 표현하고 추가된 텍스트가 없으나, 레퍼런스와 달리 비니를 착용한 점이 약간 아쉽습니다." }, { "label": "B", "score": 4, "verdict_ko": "수리영이 스마트폰을 들고 있지 않으며, 남성 직원의 외투 뒷면에 금지된 텍스트가 생성되어 지시사항을 위반했습니다." }, { "label": "C", "score": 5, "verdict_ko": "행동과 소품 상태는 대체로 정확하나, 백팩에 달린 노란색 태그에 금지된 텍스트가 포함되어 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L08B03.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S19sh5.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "수리영이 캐릭터 레퍼런스에 있는 야구 모자가 아닌 검은색 비니를 쓰고 있습니다.", "fix_en": "Change the black beanie on the woman's head to the faded baseball cap shown in the character reference." }, { "issue_ko": "수리영이 들고 있는 스마트폰의 화면이 그녀의 시선이 아닌 카메라 쪽을 향해 잘못 배치되어 있습니다.", "fix_en": "Rotate the smartphone in the woman's hand so the screen faces her body, showing the back of the phone to the camera." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the black beanie on the woman's head to the faded baseball cap shown in the character reference.\n- Rotate the smartphone in the woman's hand so the screen faces her body, showing the back of the phone 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": "플레이트+콘티+엔티티" }, "S19sh6": { "input_fingerprint": "fcc0ea2b2aca3901", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): early morning, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 수리영이 귀에 스마트폰을 댄 채 정면을 응시하는 상체 클로즈업\n\nLOCATION (lock): A large supermarket in the quiet early hours, with broad merchandise aisles, an exterior bicycle area, and outdoor walls covered by surveillance cameras. 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 exterior wall, mounted surveillance camera, and daylight from the reference. Exclude the male employee and the earlier upward-looking pose from this phone-call 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): The bicycle remains parked outside while Su-ri-young uses her smartphone to call Hye-su; her doll-adorned backpack stays with 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: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": 2 }, "selected": "B", "ranking": [ "B", "A", "C" ], "verdicts": [ { "label": "A", "score": 3, "verdict_ko": "레퍼런스의 외부 벽과 CCTV 배경을 유지해야 한다는 지시를 어기고 실내 마트 배경으로 잘못 연출되었습니다." }, { "label": "B", "score": 8, "verdict_ko": "이전 샷의 외부 벽면과 CCTV 위치를 완벽하게 유지한 채, 요구된 통화 동작과 정면 응시를 정확히 구현한 최고의 결과물입니다." }, { "label": "C", "score": 2, "verdict_ko": "외부 배경 지시를 무시하고 실내로 장소가 변경되었으며, 이전 샷과 다르게 모자를 뒤로 쓴 결정적인 형태 오류가 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L08B03.png" }, { "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/scene/recipe/S19sh5_sel.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "PROP REFERENCE — 스마트폰: the exact object appearing in this shot; match its look, material and wear exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "스마트폰이 뒤집혀 있어 화면이 귀가 아닌 카메라 쪽을 향하고 있습니다.", "fix_en": "Rotate the smartphone so its screen is pressed against her ear and its solid back casing faces the camera." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Rotate the smartphone so its screen is pressed against her ear and its solid back casing faces 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": "플레이트+prev+엔티티" }, "S20sh1": { "input_fingerprint": "b91ffcdaab563190", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): early morning, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 마트 총무과 책상 앞, 한국인 남성 보안직원과 혜수가 컴퓨터 모니터를 향해 몸을 기울인 측면\n\nLOCATION (lock): A supermarket administrative and security office with a CCTV review station, monitors, and space for staff and investigators to watch recorded footage. 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): Su-ri-young still has her doll-adorned backpack and smartphone while watching the CCTV review.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 혜수 (Korean 40대 중반 여성, 단정한 검은 머리, 성숙한 타원형 얼굴, 또렷한 이목구비); 한국인 남성 보안직원 (Korean 성인 남성, 짧고 단정한 검은 머리, 평범한 얼굴형, 깔끔한 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": 6, "C": 4 }, "selected": "A", "ranking": [ "A", "B", "C" ], "verdicts": [ { "label": "A", "score": 7, "verdict_ko": "지정된 카메라 구도와 장소 디테일을 정확히 재현했으며, 혜수의 외모, 인형 달린 가방, 스마트폰 파지 등 모든 지시를 가장 충실히 따랐습니다." }, { "label": "B", "score": 6, "verdict_ko": "구도와 장소 묘사는 우수하나, 혜수의 묶은 머리와 얼굴 형태가 캐릭터 레퍼런스의 특징과 다소 어긋납니다." }, { "label": "C", "score": 4, "verdict_ko": "장소와 인물 조건은 어느 정도 충족했으나, 레이아웃 스케치의 우측 방향 카메라 구도를 완전히 반대로 뒤집어 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L14B01.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S20sh1.png" }, { "label": "CHARACTER REFERENCE — 혜수: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "보안직원의 왼쪽 팔 끝에 오른손(엄지손가락이 카메라 쪽을 향함)이 잘못 연결되어 있습니다.", "fix_en": "Change the man's hand on the mouse to a valid left hand, with the thumb resting on the side of the mouse furthest from the camera." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the man's hand on the mouse to a valid left hand, with the thumb resting on the side of the mouse furthest from 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": "플레이트+콘티+엔티티" }, "S20sh3": { "input_fingerprint": "f7f16fdc506529b7", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): early morning, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 흑백 CCTV 화면 속 어두운 구석에 사람 형상이 희미하게 찍힌 모니터 클로즈업\n\nLOCATION (lock): A supermarket administrative and security office with a CCTV review station, monitors, and space for staff and investigators to watch recorded footage. 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 relevant section of the CCTV image remains anomalously dark, leaving the man and Min-suk only as indistinct human forms.\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": "텍스트 및 인물 배제 지침을 완벽히 준수하며 지시된 모니터 클로즈업을 사실적으로 구현함." }, { "label": "B", "score": 3, "verdict_ko": "절대 금지된 텍스트가 화면에 나타나 있고 프레임 우측에 인물의 일부가 포함되어 지침을 위반함." }, { "label": "C", "score": 4, "verdict_ko": "금지된 텍스트가 화면 상단에 삽입되었으며, 모니터가 허공에 떠 있는 듯한 비현실적인 배치가 나타남." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L14B01.png" } ], "critique": { "issues": [ { "issue_ko": "모니터가 구형 브라운관(CRT) 형태입니다. 참조된 장소 사진의 보안실에는 현대적인 평면 스크린 모니터들만 존재합니다.", "fix_en": "Change the monitor to a modern, flat-screen panel that matches the equipment shown in the location reference photograph." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change the monitor to a modern, flat-screen panel that matches the equipment shown in the location reference 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.", "ref_mode": "플레이트만 (배경 전용)" }, "S21sh7": { "input_fingerprint": "4995549e5b83a533", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 수리영이 두 눈을 크게 뜨고 입을 살짝 벌린 굳은 얼굴 클로즈업\n\nLOCATION (lock): A police station with a shared investigators’ office, clustered desks, computer monitors, paperwork, and crime-scene photographs spread across a desk. An adjacent meeting area draws the officers away. 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 enlarged murder-scene photographs remain scattered across Hye-su's desk, and Su-ri-young still has her backpack and smartphone with 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: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": 5 }, "selected": "C", "ranking": [ "C", "A", "B" ], "verdicts": [ { "label": "C", "score": 8, "verdict_ko": "요구된 표정과 소품의 방향을 잘 따랐으며, 배경 및 캐릭터 레퍼런스의 의상(올리브색 스웨터)까지 가장 정확하게 일치합니다." }, { "label": "A", "score": 6, "verdict_ko": "표정과 배경 구현은 우수하나, 캐릭터 레퍼런스와 완전히 다른 상의를 입고 있어 일관성이 떨어집니다." }, { "label": "B", "score": 5, "verdict_ko": "의상이 레퍼런스와 일치하지 않으며, 좌측 전경에 범죄 현장 사진들이 공중에 떠 있는 듯한 오류가 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L13B02.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S21sh7.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [] }, "fix_skipped": true, "ref_mode": "플레이트+콘티+엔티티" }, "S21sh10": { "input_fingerprint": "2799b91ffea8b14b", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 모니터 화면에 바다 지도가 펼쳐져 있고 붉은색 핀과 좌표 숫자가 떠 있는 클로즈업\n\nLOCATION (lock): A police station with a shared investigators’ office, clustered desks, computer monitors, paperwork, and crime-scene photographs spread across a desk. An adjacent meeting area draws the officers away. 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 monitor displays Min-suk's last phone location as offshore coordinates, with the device marked offline. Su-ri-young's backpack and smartphone remain with her.\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": "레퍼런스의 창살과 책상 소품 등 장소를 완벽하게 구현했으며, 모니터의 바다 지도, 붉은 핀, 좌표 및 오프라인 텍스트 지시를 가장 정확하게 따랐습니다." }, { "label": "B", "score": 5, "verdict_ko": "장소와 화면 구성 요구사항을 대체로 충족했으나, 모니터 인터페이스와 폴더의 텍스트가 주로 영어로 표기되어 한국적 배경의 디테일이 부족합니다." }, { "label": "C", "score": 3, "verdict_ko": "과도한 클로즈업으로 인해 지정된 경찰서 장소(레퍼런스)를 전혀 식별할 수 없으며, 요구된 텍스트 내용도 일부 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L13B02.png" } ], "critique": { "issues": [] }, "fix_skipped": true, "ref_mode": "플레이트만 (배경 전용)" }, "S21sh12": { "input_fingerprint": "9fd3ccd03b6fd940", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 수리영이 메모지가 들어간 겉옷 주머니에 손을 넣은 채 사무실 출입문을 향해 몸을 돌린 정지 순간\n\nLOCATION (lock): A police station with a shared investigators’ office, clustered desks, computer monitors, paperwork, and crime-scene photographs spread across a desk. An adjacent meeting area draws the officers away. 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): Su-ri-young carries the folded coordinate note in her pocket, with her backpack and smartphone still on her, and her bicycle waiting outside the police station.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": 3 }, "selected": "C", "ranking": [ "C", "A", "B" ], "verdicts": [ { "label": "C", "score": 7, "verdict_ko": "겉옷 주머니에 손을 넣고 출입문을 향하는 모습을 잘 구현했으며, 밖에 있어야 할 자전거가 실내에 나타나지 않았습니다." }, { "label": "A", "score": 4, "verdict_ko": "겉옷이 아닌 바지 뒷주머니에 손을 넣었으며, 밖에 대기해야 할 자전거가 실내에 잘못 배치되었습니다." }, { "label": "B", "score": 3, "verdict_ko": "주머니에 넣은 손 외에 스마트폰을 들고 있는 세 번째 손이 나타나는 심각한 신체 생성 오류가 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L13B02.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S21sh12.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "스마트폰의 화면(기능적인 면)이 카메라를 향해 노출되어 있어, 지시문에서 특별히 요구하지 않는 한 기능적 면을 카메라에 보이지 않도록 하는 규칙을 위반했습니다.", "fix_en": "Rotate the smartphone in the right hand so the screen faces the character's body, making only the back of the phone visible to the camera." }, { "issue_ko": "스마트폰을 쥐고 있는 오른손의 손가락들이 화면과 융합되고 형태가 일그러져 해부학적으로 올바르지 않습니다.", "fix_en": "Redraw the right hand holding the smartphone to have clear, anatomically correct fingers." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Rotate the smartphone in the right hand so the screen faces the character's body, making only the back of the phone visible to the camera.\n- Redraw the right hand holding the smartphone to have clear, anatomically correct 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": "플레이트+콘티+엔티티" }, "S22sh1": { "input_fingerprint": "69f49eb87cbbd75a", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day.\n\nSHOT TEXT (authoritative, Korean): 경찰서 건물 밖 도로 위에서 수리영이 자전거 안장에 앉아 상체를 앞으로 숙인 채 머리카락이 뒤로 흩날리는 측면\n\nLOCATION (lock): The exterior of a police station with a bicycle parked nearby and direct access to the street. 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): Su-ri-young rides away on the bicycle while retaining the folded coordinate note, her doll-adorned backpack, and her smartphone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": "배경과 인물의 자세를 잘 살렸으며, 모든 소지품(스마트폰, 인형, 접힌 쪽지)을 해부학적 오류 없이 가장 자연스럽게 배치했습니다." }, { "label": "A", "score": 5, "verdict_ko": "전반적인 지시를 따랐으나 접힌 쪽지가 소매에 부자연스럽게 붙어 있어 현실감이 다소 떨어집니다." }, { "label": "B", "score": 3, "verdict_ko": "휴대폰을 들고 있는 손의 해부학적 구조가 심하게 왜곡되어 사실적인 연출에 실패했습니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L15B01.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S22sh1.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "캐릭터가 손에 쥐고 있는 종이에 글씨나 낙서가 적혀 있습니다 (프롬프트의 텍스트 금지 조항 위반).", "fix_en": "Remove all text, writing, and scribbles from the folded paper in the character's hand, making it entirely blank." }, { "issue_ko": "오른쪽 핸들을 잡은 손과 종이를 쥐고 있는 손가락이 기형적으로 융합되어 손가락이 비정상적으로 많습니다.", "fix_en": "Correct the right hand's anatomy to be a single, anatomically correct hand with five fingers gripping both the handlebar and the paper." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all text, writing, and scribbles from the folded paper in the character's hand, making it entirely blank.\n- Correct the right hand's anatomy to be a single, anatomically correct hand with five fingers gripping both the handlebar and the paper.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S23sh1": { "input_fingerprint": "344ce70c0eb8cb8a", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 어두운 PC방 안, 모니터 불빛이 비치는 수리영의 정면 상체\n\nLOCATION (lock): A computer café with rows of monitor stations and access to a printer for producing multiple map sheets. 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): Su-ri-young has the folded coordinate note available while looking up the location, and her backpack and smartphone remain with 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: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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, "A": 5, "B": 3 }, "selected": "C", "ranking": [ "C", "A", "B" ], "verdicts": [ { "label": "C", "score": 6, "verdict_ko": "'정면 상체' 구도와 조명, 프린터 위치는 정확하나, 메모지 방향이 카메라를 향해 틀렸고 스마트폰이 누락되었으며 미지정 인물과 글자가 포함되었습니다." }, { "label": "A", "score": 5, "verdict_ko": "소지품 상태와 장소 디테일은 양호하지만, 가장 중요한 '정면' 구도를 무시한 측면 샷이며 미지정 배경 인물과 메모지 내 글자가 포함되었습니다." }, { "label": "B", "score": 3, "verdict_ko": "정면 구도에 가깝지만 모니터와 파티션 구조가 비현실적으로 왜곡되었고, 필수 배경인 프린터가 누락되었으며 미지정 인물이 등장합니다." } ], "refs": [ { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "샷 텍스트에 지정되지 않은 다른 사람들이 배경에 다수 포함되어 있음.", "fix_en": "Remove all background people from the shot so that only Su-ri-young is visible." }, { "issue_ko": "화면 우측 모니터에 붙은 종이에 금지된 텍스트(글씨)가 적혀 있음.", "fix_en": "Remove all visible text and scribbles from the paper attached to the monitor." }, { "issue_ko": "수리영이 참고해야 할 좌표 메모지가 그녀의 시선 반대편인 모니터 뒷면에 카메라를 향해 붙어 있어 물리적 방향이 어긋남.", "fix_en": "Reposition the coordinate note so it is placed on the desk facing Su-ri-young's eyes, not facing the camera." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove all background people from the shot so that only Su-ri-young is visible.\n- Remove all visible text and scribbles from the paper attached to the monitor.\n- Reposition the coordinate note so it is placed on the desk facing Su-ri-young's eyes, not facing 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": "플레이트+콘티+엔티티" }, "S23sh4": { "input_fingerprint": "eca5130d4844e97c", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 수리영의 두 손이 접힌 지도 종이들을 열려 있는 백팩 안으로 밀어넣은 정지 순간\n\nLOCATION (lock): A computer café with rows of monitor stations and access to a printer for producing multiple map sheets. 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 dark computer-booth lighting, monitor glow, and workstation materials from the reference. Exclude the seated upper body and screen content, focusing only on the hands, folded maps, and open 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): Several folded printed maps are now stored inside Su-ri-young's doll-adorned backpack; she also retains the coordinate note and smartphone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": "상체를 제외하고 손과 가방에만 집중하라는 프레이밍 지시를 정확히 따랐으며, 지도 소품의 디테일과 PC방 배경이 참조 이미지와 훌륭하게 일치합니다." }, { "label": "B", "score": 4, "verdict_ko": "지시된 프레이밍은 어느 정도 준수했으나 팔의 방향이 부자연스럽고 지도 측면에 참조에 없는 불필요한 텍스트가 추가되었습니다." }, { "label": "C", "score": 3, "verdict_ko": "손과 가방, 지도에만 초점을 맞추라는 지시를 어기고 인물의 상반신이 넓게 프레임에 포함되어 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/scene/recipe/S23sh1_sel.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "PROP REFERENCE — 인쇄된 해안가 지도: the exact object appearing in this shot; match its look, material and wear exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "아래쪽 손에 5개의 손가락 외에 가방 쪽으로 길게 뻗은 불필요한 여섯 번째 손가락이 있습니다.", "fix_en": "Remove the extra, bottom-most downward-pointing finger from the lower hand to leave a normal five-digit hand." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the extra, bottom-most downward-pointing finger from the lower hand to leave a normal five-digit 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": "플레이트+prev+엔티티" }, "S24sh2": { "input_fingerprint": "51a94ccea0fe2c13", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): day, heavy thunderstorm.\n\nSHOT TEXT (authoritative, Korean): 비에 흠뻑 젖은 수리영이 자전거 핸들을 꽉 쥔 채 앞을 노려보는 얼굴 클로즈업\n\nLOCATION (lock): A coastal route exposed to a sudden downpour, with dark storm clouds, thunder, lightning, and heavy rain over the bicycle 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\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Su-ri-young rides through the rain with the printed maps secured in her backpack, while retaining the coordinate note and smartphone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": "레퍼런스와 의상이 정확히 일치하며 폭우 속 자전거를 타는 모습을 사실적으로 구현했습니다." }, { "label": "A", "score": 6, "verdict_ko": "전반적으로 잘 일치하나 자전거 핸들과 브레이크 형태에 시각적 오류가 있습니다." }, { "label": "C", "score": 4, "verdict_ko": "프롬프트와 레퍼런스에 없는 후드 재킷을 임의로 추가하여 의상 일관성을 위반했습니다." } ], "refs": [ { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "배낭을 메고 있어야 하나 어깨에 배낭 끈이 전혀 보이지 않습니다.", "fix_en": "Add visible backpack straps over her shoulders." }, { "issue_ko": "유지하고 있어야 할 좌표 메모와 스마트폰이 화면에 보이지 않습니다.", "fix_en": "Add the coordinate note and smartphone into one of her hands gripping the handlebar." }, { "issue_ko": "손이 자전거 핸들의 끝부분 고무 그립이 아닌 안쪽 금속 바를 잡고 있어 쥐는 위치가 부자연스럽습니다.", "fix_en": "Reposition her hands outward so they are holding the textured grips at the ends of the handlebars." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add visible backpack straps over her shoulders.\n- Add the coordinate note and smartphone into one of her hands gripping the handlebar.\n- Reposition her hands outward so they are holding the textured grips at the ends of the handlebars.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S25sh1": { "input_fingerprint": "da98967b3541c22c", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk, open window.\n\nSHOT TEXT (authoritative, Korean): 해질 무렵 옥탑방 거실, 흙 묻은 발자국들이 찍힌 바닥과 쓰러진 식탁 의자들이 보이는 전경\n\nLOCATION (lock): A ransacked rooftop apartment with muddy or dusty footprints, an open living-room window, overturned dining chairs, and household items pulled apart. The bedroom contains a red circle on the wall with a bloodstained photograph pinned at its center. 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 apartment has been ransacked, with muddy footprints, displaced household items, toppled chairs, and an open window. The bedroom already contains a red wall circle with the bloodied old photograph pinned at its center; Su-ri-young still has the maps in her backpack.\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": 6, "B": 3 }, "selected": "C", "ranking": [ "C", "A", "B" ], "verdicts": [ { "label": "C", "score": 7, "verdict_ko": "원본의 공간 요소를 유지하며 발자국, 열린 창문, 침실의 붉은 원과 사진 등 요구된 난장판 상태를 가장 정확하게 구현했습니다." }, { "label": "A", "score": 6, "verdict_ko": "대부분의 지시를 따랐으나 원본 위치에 있던 식탁이 사라져 공간의 일관성이 다소 떨어집니다." }, { "label": "B", "score": 3, "verdict_ko": "침실의 침대와 필수 요소인 붉은 원, 사진이 완전히 누락되어 공간 요구사항을 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B03.png" } ], "critique": { "issues": [ { "issue_ko": "거실 창문이 참고 이미지의 고정된 건축적 특징인 미닫이창이 아닌 여닫이창으로 잘못 렌더링되었습니다.", "fix_en": "Render the open living room window as a sliding window rather than swinging casement windows, matching the reference architecture." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Render the open living room window as a sliding window rather than swinging casement windows, matching the reference architecture.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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)" }, "S25sh5": { "input_fingerprint": "7166b60a531e04ac", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk, open window.\n\nSHOT TEXT (authoritative, Korean): 붉은 원 한가운데 피 묻은 낡은 흑백 사진이 벽에 꽂혀 있는 타이트 샷\n\nLOCATION (lock): A ransacked rooftop apartment with muddy or dusty footprints, an open living-room window, overturned dining chairs, and household items pulled apart. The bedroom contains a red circle on the wall with a bloodstained photograph pinned at its center. 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 ransacked apartment and muddy footprints remain undisturbed, and the bloodied old photograph is still pinned at the center of the red circle.\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": 6, "B": 4 }, "selected": "C", "ranking": [ "C", "A", "B" ], "verdicts": [ { "label": "C", "score": 7, "verdict_ko": "타이트 샷 요구를 정확히 충족하며 낡은 흑백 사진 속 인물과 피자국, 벽면 디테일을 가장 사실적으로 재현했습니다." }, { "label": "A", "score": 6, "verdict_ko": "타이트 샷 지시는 따랐으나 사진 속 피사체가 다소 모호하고 압정 등의 디테일이 약간 아쉽습니다." }, { "label": "B", "score": 4, "verdict_ko": "타이트 샷 지시를 무시하고 침대와 창문이 포함된 넓은 앵글을 잡아 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B02.png" } ], "critique": { "issues": [] }, "fix_skipped": true, "ref_mode": "플레이트만 (배경 전용)" }, "S25sh9": { "input_fingerprint": "0e8e155d18cbc857", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk, open window.\n\nSHOT TEXT (authoritative, Korean): 수리영의 손이 구겨진 사진을 꽉 움켜쥔 타이트 샷\n\nLOCATION (lock): A ransacked rooftop apartment with muddy or dusty footprints, an open living-room window, overturned dining chairs, and household items pulled apart. The bedroom contains a red circle on the wall with a bloodstained photograph pinned at its center. 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 photograph has been removed from the wall and is now tightly clenched in Su-ri-young's hand, while the red circle and the ransacked state of the apartment remain behind. Her backpack still contains the printed maps.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": 6, "C": 2 }, "selected": "A", "ranking": [ "A", "B", "C" ], "verdicts": [ { "label": "A", "score": 7, "verdict_ko": "해질녘 조명과 구겨진 사진 소품, 난장판이 된 배경은 훌륭하게 구현되었으나 캐릭터의 긴팔 의상이 맨팔로 표현된 점이 아쉽습니다." }, { "label": "B", "score": 6, "verdict_ko": "배경과 소품 묘사는 무난하지만 해질녘의 빛 느낌이 다소 부족하며, A와 마찬가지로 캐릭터의 소매 의상이 누락되었습니다." }, { "label": "C", "score": 2, "verdict_ko": "잘린 팔이 다른 팔에 어색하게 겹쳐 있는 등 해부학적 구조와 형태가 심각하게 붕괴된 치명적인 오류가 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B02.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S25sh9.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "PROP REFERENCE — 구겨진 피 묻은 낡은 흑백 사진: the exact object appearing in this shot; match its look, material and wear exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "손의 해부학적 구조가 잘못되었습니다 (왼팔 시점임에도 엄지손가락이 왼쪽에 있고 손가락 관절이 기형적으로 꺾여 있습니다).", "fix_en": "Re-render the hand as an anatomically correct left hand with the thumb on the right side, naturally gripping the photograph." }, { "issue_ko": "프롬프트에 명시된 '열린 창문(open window)' 지시와 달리 우측의 창문이 닫혀 있습니다.", "fix_en": "Change the window on the right side so that it is clearly open." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Re-render the hand as an anatomically correct left hand with the thumb on the right side, naturally gripping the photograph.\n- Change the window on the right side so that it is clearly open.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S26sh2": { "input_fingerprint": "4b8e061ee2695060", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk, dense fog after rain.\n\nSHOT TEXT (authoritative, Korean): 수리영이 갯바위 위에 무릎을 모으고 앉아 바다를 응시하는 측면\n\nLOCATION (lock): An empty harbor after a downpour, bordered by tidal rocks and filled with dense fog that hides the boats offshore. A single small fishing boat approaches through 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\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Su-ri-young continues to possess the bloodied old photograph and carries the printed maps in her backpack, along with her smartphone.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": 7, "C": 5 }, "selected": "B", "ranking": [ "B", "A", "C" ], "verdicts": [ { "label": "B", "score": 8, "verdict_ko": "갯바위 위에 앉은 완벽한 측면 자세와 배낭 안의 지도를 정확히 묘사했으나 레퍼런스의 모자가 누락되었습니다." }, { "label": "A", "score": 7, "verdict_ko": "모자를 착용하고 부두 옆 바위에 앉은 모습은 좋으나 지시된 지도가 배낭에서 확인되지 않습니다." }, { "label": "C", "score": 5, "verdict_ko": "모자를 쓰고 바위에 앉아 있으나 배낭을 벗어두어 소지 상태(carried state)를 위반했으며 지도도 보이지 않습니다." } ], "refs": [ { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "캐릭터가 참조 이미지에 있는 회색 모자를 쓰고 있지 않습니다.", "fix_en": "Add the grey cap from the reference image to the character's head." }, { "issue_ko": "오른쪽 손목에 시계가 차여 있습니다 (참조 이미지에서는 왼쪽 손목).", "fix_en": "Move the watch from the right wrist to the left wrist." }, { "issue_ko": "무릎에 얹은 오른팔 끝에 왼손이 달려 있으며 손가락 형태가 뭉개져 있습니다.", "fix_en": "Fix the hand resting on the knee so it is an anatomically correct right hand." }, { "issue_ko": "아래쪽에 위치한 신발이 다리와 연결되지 않은 채 허공에 떠 있고, 다리 해부학이 엉켜 있습니다.", "fix_en": "Remove the disconnected lower shoe and fix the legs to show only naturally connected feet." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the grey cap from the reference image to the character's head.\n- Move the watch from the right wrist to the left wrist.\n- Fix the hand resting on the knee so it is an anatomically correct right hand.\n- Remove the disconnected lower shoe and fix the legs to show only naturally connected feet.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S26sh5": { "input_fingerprint": "21f116115f4757c4", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk, dense fog after rain.\n\nSHOT TEXT (authoritative, Korean): 뱃머리에서 내린 인우가 굵은 닻줄을 쥔 채 바닥의 말뚝에 걸친 정지 순간\n\nLOCATION (lock): An empty harbor after a downpour, bordered by tidal rocks and filled with dense fog that hides the boats offshore. A single small fishing boat approaches through 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\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): In-woo is securing the boat with the mooring line while Su-ri-young retains the old photograph, her map-filled backpack, and her smartphone nearby.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 인우 (Korean 20대 중반 남성, 자연스러운 검은 머리, 균형 잡힌 젊은 얼굴, 선이 고운 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": 4, "A": 3 }, "selected": "B", "ranking": [ "B", "C", "A" ], "verdicts": [ { "label": "B", "score": 7, "verdict_ko": "샷 텍스트에 명시되지 않은 인물을 엄격히 제외하는 규칙을 유일하게 준수했으며, 인우의 착장과 안개 낀 항구 배경을 충실하게 구현했습니다." }, { "label": "C", "score": 4, "verdict_ko": "인우의 닻줄 고정 묘사는 우수하나, 샷 텍스트에 없는 인물(수리영)을 화면에 포함시켜 인물 제한 규칙을 위반했습니다." }, { "label": "A", "score": 3, "verdict_ko": "샷 텍스트에 허용되지 않은 인물을 등장시켰고, 인우가 밧줄을 다루는 자세가 요구된 정지 순간의 묘사로 보기 어렵습니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/map_plate_S26sh5.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S26sh5.png" }, { "label": "CHARACTER REFERENCE — 인우: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "인우가 서서 밧줄을 당기는 레이아웃 스케치의 자세와 달리 부두에 완전히 쪼그려 앉아 있습니다.", "fix_en": "Change In-woo's pose to standing with legs apart and leaning back to pull the rope, matching the layout sketch." }, { "issue_ko": "인우의 오른손 손가락 사이로 밧줄이 관통하여 녹아내렸고, 왼손은 형태가 뭉개져 밧줄을 쥐고 있지 않습니다.", "fix_en": "Redraw In-woo's hands with correct anatomy so his fingers firmly and realistically wrap around the rope without blending into it." }, { "issue_ko": "'텅 빈 항구'라는 위치 지시어와 달리 부두 왼쪽에 선명하게 정박된 어선들이 추가되어 있습니다.", "fix_en": "Remove the docked fishing boats on the left side of the pier to make the harbor empty, as specified in the location prompt." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Change In-woo's pose to standing with legs apart and leaning back to pull the rope, matching the layout sketch.\n- Redraw In-woo's hands with correct anatomy so his fingers firmly and realistically wrap around the rope without blending into it.\n- Remove the docked fishing boats on the left side of the pier to make the harbor empty, as specified in the location prompt.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S26sh9": { "input_fingerprint": "c88360442c3fdf88", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk, dense fog after rain.\n\nSHOT TEXT (authoritative, Korean): 수리영이 바다를 향해 선 채 스마트폰을 입가에 댄 측면 상체\n\nLOCATION (lock): An empty harbor after a downpour, bordered by tidal rocks and filled with dense fog that hides the boats offshore. A single small fishing boat approaches through the mist. 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 foggy harbor, mooring area, wet coastal textures, and dusk light from the reference. Exclude the man, the rope-handling action, and the boat's disembarkation moment from this phone-message 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): Su-ri-young records the message on her smartphone while still carrying the old photograph and the backpack containing her printed maps.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": "스마트폰을 입가에 댄 자세와 소지품(사진, 배낭) 지시를 가장 잘 구현했으나, 화면에 미세한 텍스트가 보입니다." }, { "label": "B", "score": 5, "verdict_ko": "지도가 배낭 안에 있어야 한다는 지시를 어기고 손에 들고 있으며, 머리 모양이 묘사와 다릅니다." }, { "label": "C", "score": 4, "verdict_ko": "스마트폰을 입가에 대라는 핵심 지시를 어기고 귀에 댔으며, 폰의 외형도 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/scene/recipe/S26sh5_sel.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "PROP REFERENCE — 스마트폰: the exact object appearing in this shot; match its look, material and wear exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "스마트폰의 화면이 캐릭터의 시선이 아닌 카메라를 향해 부자연스럽게 노출되어 있습니다.", "fix_en": "Rotate the smartphone so the screen faces the character's eyes, showing its back or side edge to the camera." }, { "issue_ko": "스마트폰 화면에 텍스트와 UI가 렌더링되어 '텍스트 없음' 지시를 위반했습니다.", "fix_en": "Remove all text, numbers, and UI elements from the smartphone screen." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Rotate the smartphone so the screen faces the character's eyes, showing its back or side edge to the camera.\n- Remove all text, numbers, and UI elements from the smartphone screen.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+prev+엔티티" }, "S27sh3": { "input_fingerprint": "70d75d3e1065af60", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk, interior lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 혜수가 미간을 깊게 찌푸린 채 스마트폰을 내려다보는 얼굴 클로즈업\n\nLOCATION (lock): The interior of a small rooftop apartment, where a recorded phone message is being checked. 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 disturbed living room, muddy footprints, overturned dining chairs, open window, and dusk lighting from the reference. Exclude the younger woman and her search action, showing only the officer checking the phone while the disorder remains.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\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 apartment remains ransacked, with muddy footprints, toppled chairs, an open window, and the red wall circle still present; the pinned photograph is gone because Su-ri-young took it. Hye-su is holding the smartphone on which she is checking Su-ri-young's message.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 혜수 (Korean 40대 중반 여성, 단정한 검은 머리, 성숙한 타원형 얼굴, 또렷한 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": "화면 하단에 자막이 생성되어 텍스트 배제 규칙을 위반했으나, 기준 공간의 세부 구조, 캐릭터의 지정 의상, 소품 방향 등 나머지 모든 지시를 가장 정확히 따랐습니다." }, { "label": "B", "score": 3, "verdict_ko": "캐릭터 의상과 표정은 훌륭하나, 방의 좌우 구조가 뒤집히고 주방 등 핵심 요소가 심하게 왜곡되어 공간 일치(Location Lock) 규칙을 크게 훼손했습니다." }, { "label": "C", "score": 4, "verdict_ko": "기준 공간은 잘 재현했으나, 캐릭터가 지정된 사복 대신 경찰 제복을 입었고 스마트폰 화면이 카메라 쪽을 향해 소품 방향 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L04B03.png" }, { "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/scene/recipe/S25sh1_sel.png" }, { "label": "CHARACTER REFERENCE — 혜수: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "이미지 하단에 자막 텍스트가 생성되어 '텍스트 없음' 지침을 위반했습니다.", "fix_en": "Remove the Korean subtitle text entirely from the bottom of the image." }, { "issue_ko": "배경 침실의 붉은 원 안에 사진이 그대로 있습니다. 프롬프트는 수리영이 사진을 가져갔기 때문에 없어져야 한다고 명시했습니다.", "fix_en": "Remove the small photograph pinned to the wall inside the red circle in the background." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the Korean subtitle text entirely from the bottom of the image.\n- Remove the small photograph pinned to the wall inside the red circle 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": "플레이트+prev+엔티티" }, "S28sh5": { "input_fingerprint": "25853cfd9bc2472b", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 인우가 사진에 시선을 고정한 채 두 눈을 크게 뜨고 상체를 뒤로 뺀 측면 구도\n\nLOCATION (lock): A small fishing boat far out at sea, with an open working deck and an enclosed wheelhouse. The surrounding water stretches away from the harbor. 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): Su-ri-young has brought the bloodied old photograph aboard and is showing it to In-woo; her backpack still holds the remaining printed maps, while one map has been taken out for navigation.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 인우 (Korean 20대 중반 남성, 자연스러운 검은 머리, 균형 잡힌 젊은 얼굴, 선이 고운 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": 6, "B": 4, "C": 3 }, "selected": "A", "ranking": [ "A", "B", "C" ], "verdicts": [ { "label": "A", "score": 6, "verdict_ko": "샷 텍스트의 '크게 뜬 눈'과 '뒤로 뺀 상체'를 잘 구현했으나, 지문 외 인물 등장 및 사진이 반대로 향한 점이 아쉽습니다." }, { "label": "B", "score": 4, "verdict_ko": "인우의 눈을 크게 뜬 표정 묘사가 부족하며, 화면 좌측에 부자연스러운 팔이 등장하고 소품 방향 규정을 위반했습니다." }, { "label": "C", "score": 3, "verdict_ko": "인우의 하반신이 갑판 바닥에 파묻히는 심각한 렌더링 오류가 있어 실사 사진으로서의 사실성이 훼손되었습니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L19B01.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S28sh5.png" }, { "label": "CHARACTER REFERENCE — 인우: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "샷 텍스트에 지시되지 않은 인물(여성)이 프레임에 포함됨", "fix_en": "Remove the female character completely, as the shot text only permits In-woo to be visible." }, { "issue_ko": "인우에게 보여주는 사진의 앞면(이미지)이 인우의 시선이 아닌 카메라 쪽을 향하고 있음", "fix_en": "Flip the photograph so its functional side (the bloody image) faces In-woo's eyes." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Remove the female character completely, as the shot text only permits In-woo to be visible.\n- Flip the photograph so its functional side (the bloody image) faces In-woo's 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": "플레이트+콘티+엔티티" }, "S28sh8": { "input_fingerprint": "9284ac0a8ba8b555", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk.\n\nSHOT TEXT (authoritative, Korean): 수리영이 조타실 입구에 서서 두 팔을 벌린 채 멈춘 전신\n\nLOCATION (lock): A small fishing boat far out at sea, with an open working deck and an enclosed wheelhouse. The surrounding water stretches away from the 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\nPREVIOUS STILL USAGE (follow exactly — what to take from the attached still and what to exclude): Take the small fishing boat's deck materials, dusk light, surrounding sea, and wheelhouse exterior from the reference. Exclude the man's startled seated reaction and the close presentation of the photograph, framing the woman alone at the wheelhouse entrance.\n\nREALIZE FIGURATIVE LANGUAGE AS A LIVE-ACTION SHOT: the Korean shot\ntext may describe characters metaphorically, figuratively or with\nexaggeration. Photograph what a real movie camera would actually\nrecord on a physical set — exaggerated or figurative impressions\nbecome realistic staging choices (lighting, distance, angle,\nwardrobe), not literal fantasy imagery.\nEVERY CHARACTER IS A HUMAN BEING: unless the story explicitly\nfeatures non-human or virtual beings (as in science fiction or\nfantasy), every character — however indirectly, vaguely or\nfiguratively the text describes them — IS a real human. When the\ntext gives no direct visual description of a person, IMAGINE one\nand still show them as a concrete, fully-formed human being:\nalways render the human form to the maximum extent — build,\nposture, face, hands, clothing — never reduce a person to a shape,\nblob, solid silhouette or abstract mass.\n\nEXPRESSIONS ARE ACTED, NEVER ANATOMICAL: when the text describes a\nperson's eyes, face or presence emotionally or figuratively (vacant,\nhollow, dazed, burning, lifeless gaze and the like), it describes an\nACTOR'S PERFORMANCE captured by a real camera — realize it ONLY\nthrough gaze direction, focus, eyelids, facial muscles, stillness\nand posture. Every living person's eyes remain anatomically normal\nhuman eyes with a natural iris and pupil, natural sclera and normal\nproportions; NEVER whiten, blank out, cloud over, glow, enlarge or\notherwise alter eyeballs, skin or anatomy — unless the story\nexplicitly declares that being non-human or supernatural in form.\n\nPROPS FACE THE RIGHT WAY: every handheld or used object must be\noriented exactly as its real-world use requires. A person reading,\nwatching or operating something (a phone, a photograph, a paper,\nany device) has its functional side — screen, front, page — facing\nTHEIR OWN eyes; the camera then sees whatever side the staging\ngeometry implies (often its back). Show the functional side to the\ncamera ONLY when the shot text itself stages it toward the viewer.\nNever flip, mirror or reverse an object's front and back.\n\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): Su-ri-young retains the old photograph and her map-filled backpack as she blocks the wheelhouse entrance. The many circles inside the wheelhouse, including the red one revealed next, must already be physically present.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": "배경과 두 팔을 벌린 전신 자세는 기준에 부합하나, 손에 들고 있어야 할 낡은 사진 소품이 누락되었습니다." }, { "label": "B", "score": 7, "verdict_ko": "지시된 소품(사진, 배낭)과 자세를 가장 충실히 재현했으나, 배낭 안의 지도가 시각적으로 뚜렷하지 않습니다." }, { "label": "C", "score": 3, "verdict_ko": "소품은 포함되었으나 캐릭터의 모자가 빠졌고, 조타실의 구조와 문에 왜곡된 원형 무늬가 추가되어 장소 일관성이 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L19B02.png" }, { "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/scene/recipe/S28sh5_sel.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "이전 장면에 있던 바닥의 피 묻은 지도와 밧줄이 사라짐", "fix_en": "Restore the bloody map and coiled rope on the deck between the hatch and the wheelhouse entrance." }, { "issue_ko": "가방에 지도가 들어있다는 설정이 보이지 않음", "fix_en": "Show the rolled maps protruding from the top of the backpack she is wearing, matching its previous state." }, { "issue_ko": "들고 있는 사진의 앞면이 부자연스럽게 카메라를 향함", "fix_en": "Rotate the photograph in her hand so its blank back faces the camera, adhering to the prop orientation rule." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Restore the bloody map and coiled rope on the deck between the hatch and the wheelhouse entrance.\n- Show the rolled maps protruding from the top of the backpack she is wearing, matching its previous state.\n- Rotate the photograph in her hand so its blank back faces the camera, adhering to the prop orientation rule.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+prev+엔티티" }, "S29sh3": { "input_fingerprint": "d2ae606db9597486", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk, wheelhouse lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 수리영의 검지손가락이 낡은 사진 속 아이의 얼굴 부분에 닿아 있는 클로즈업\n\nLOCATION (lock): A compact fishing-boat wheelhouse with steering controls and interior surfaces marked by numerous circular forms, including a distinct red circle. 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 old photograph remains in Su-ri-young's possession as her finger indicates the child. Numerous circles, including a red circle, remain marked throughout the wheelhouse behind In-woo.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태) — never anyone else, and never add a person the shot text does not show. IF only part of a person is in frame (a hand, arm, foot, back, silhouette), that body part belongs to the specific person the shot text 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": 5 }, "selected": "B", "ranking": [ "B", "C", "A" ], "verdicts": [ { "label": "A", "score": 3, "verdict_ko": "지시되지 않은 인물의 측면이 우측 화면에 등장하며, 허공에 뜬 주석 형태의 빨간 원이 포함되어 규정을 위반했습니다." }, { "label": "B", "score": 7, "verdict_ko": "사진 속 아이의 얼굴을 가리키는 핵심 동작을 유일하게 구현했으며, 조타실 내부에 빨간 원을 자연스러운 사물로 잘 배치했습니다." }, { "label": "C", "score": 5, "verdict_ko": "기준 소품의 배경은 잘 재현했으나 사진 속 아이들이 생략되어 지시된 동작(아이의 얼굴에 닿는 손가락)을 표현하지 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L20B01.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S29sh3.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "PROP REFERENCE — 구겨진 피 묻은 낡은 흑백 사진: the exact object appearing in this shot; match its look, material and wear exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "손에 들고 있는 사진 속 이미지가 제공된 레퍼런스(두 소녀가 벤치에 앉아 있는 모습)와 전혀 다르게 한 인물의 얼굴 클로즈업으로 변형되어 생성됨.", "fix_en": "Replace the image on the held photograph to exactly match the provided prop reference, showing two girls seated full-body on a bench." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Replace the image on the held photograph to exactly match the provided prop reference, showing two girls seated full-body on a bench.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S29sh4": { "input_fingerprint": "1b0688328d33fbae", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk, wheelhouse lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 인우가 입술을 살짝 벌린 채 하얗게 질린 얼굴 타이트 샷\n\nLOCATION (lock): A compact fishing-boat wheelhouse with steering controls and interior surfaces marked by numerous circular forms, including a distinct red circle. 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 wheelhouse remains covered with numerous circles, including the red circle, while Su-ri-young retains the old photograph and her backpack of maps.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 인우 (Korean 20대 중반 남성, 자연스러운 검은 머리, 균형 잡힌 젊은 얼굴, 선이 고운 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": "인우의 의상과 창백한 표정을 잘 표현했으며, 지시대로 빨간 원을 유지하고 책상 위 사진을 올바르게 제거했습니다." }, { "label": "B", "score": 3, "verdict_ko": "빨간 원이 누락되었고, 수리영이 소지해야 할 사진과 배낭이 책상에 있으며, 인우의 의상도 일치하지 않습니다." }, { "label": "C", "score": 5, "verdict_ko": "빨간 원과 인물 묘사는 우수하나, 수리영이 가지고 있어야 할 옛날 사진이 여전히 책상에 남아 있어 유지 상태 설정에 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L20B01.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S29sh4.png" }, { "label": "CHARACTER REFERENCE — 인우: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "프롬프트가 요구한 '붉은색 원'이 물리적인 공간 표면에 있는 것이 아니라, 공중에 떠 있는 붉은색 선(디지털 주석)처럼 묘사되어 '주석 금지(no annotations)' 지시를 위반했습니다.", "fix_en": "Render the red circle as a physical painted mark on the interior surfaces, rather than a floating digital annotation line." }, { "issue_ko": "로케이션 레퍼런스 사진 속 왼쪽 데스크 위에 있던 물병, 작업용 장갑, 테이프 등의 소품들이 모두 누락되었습니다.", "fix_en": "Restore the water bottle, work gloves, and tape rolls to the desk on the left, matching the location reference photograph." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Render the red circle as a physical painted mark on the interior surfaces, rather than a floating digital annotation line.\n- Restore the water bottle, work gloves, and tape rolls to the desk on the left, matching the location reference 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.", "ref_mode": "플레이트+콘티+엔티티" }, "S29sh8": { "input_fingerprint": "bd8c8027f8d193bc", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): dusk, wheelhouse lighting unspecified.\n\nSHOT TEXT (authoritative, Korean): 눈이 반쯤 감긴 수리영의 이마가 인우의 어깨에 맞닿아 있는 찰나\n\nLOCATION (lock): A compact fishing-boat wheelhouse with steering controls and interior surfaces marked by numerous circular forms, including a distinct red circle. 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): Suriyoung is collapsed motionless against Inwoo’s shoulder as her consciousness fades. The source does not specify her facing direction, exact head and torso angles, arm positions, leg positions, or other supporting contact.\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 wheelhouse circles remain in place as Suriyoung is collapsed motionless against Inwoo’s shoulder as her consciousness fades. Her old photograph, backpack, maps, and smartphone remain aboard with her, though the photograph's exact grip after collapse is unspecified.\n\nPEOPLE: the SHOT TEXT alone decides whether any person is visible in this shot. IF a person appears, they must be one of: 수리영 (Korean 20대 초반 여성, 긴 검은 머리, 타원형의 젊은 얼굴, 부드러운 이목구비와 자연스러운 피부색상태); 인우 (Korean 20대 중반 남성, 자연스러운 검은 머리, 균형 잡힌 젊은 얼굴, 선이 고운 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": "눈이 반쯤 감긴 수리영의 표정과 중력에 순응하는 무기력한 자세를 지시문대로 가장 정확하게 연출했습니다." }, { "label": "B", "score": 5, "verdict_ko": "수리영의 눈이 완전히 감겨 있어 '반쯤 감긴'이라는 텍스트와 불일치하며, 인우가 팔로 안는 능동적인 포즈가 추가되었습니다." }, { "label": "A", "score": 3, "verdict_ko": "수리영의 눈이 완전히 감겨 있을 뿐만 아니라, 위치 레퍼런스의 마크업인 빨간 원을 허공에 뜬 실물로 렌더링하는 치명적인 오류를 범했습니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L20B01.png" }, { "label": "LAYOUT SKETCH — a bare thin-line layout guide, a REFERENCE ONLY: take from it ONLY the camera framing, figure placement, pose and size/depth order. It carries ZERO visual style — every texture, material, light and all realism come from the text and the photographic reference. Never let any line-drawing quality leak into the output.", "path": "/Users/manta/Documents/Projects/TheRoad-I1/projects/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/images/e666c62b-be34-4abd-9c4d-15e4eb753eba/conti/conti_S29sh8.png" }, { "label": "CHARACTER REFERENCE — 수리영: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" }, { "label": "CHARACTER REFERENCE — 인우: the exact person appearing in this shot; match face, hair and build exactly.", "path": "" } ], "critique": { "issues": [ { "issue_ko": "수리영의 눈이 완전히 감겨 있어 '눈이 반쯤 감긴'이라는 텍스트 지문을 위반했습니다.", "fix_en": "Open Suriyoung's eyelids halfway." }, { "issue_ko": "낡은 사진이 책상 위와 수리영의 오른손에 각각 하나씩 중복되어 생성되었습니다.", "fix_en": "Remove the duplicated small photograph from Suriyoung's right hand." }, { "issue_ko": "의식을 잃은 수리영의 오른손이 근육에 힘을 주어 사진을 꼬집듯이 쥐고 있어, 무근육(no muscular effort) 지문을 위반했습니다.", "fix_en": "Relax Suriyoung's right fingers completely so they rest limply without pinching." }, { "issue_ko": "프롬프트 로케이션에 명시된 '뚜렷한 빨간색 원(distinct red circle)'이 조타실 내부 표면에 존재하지 않습니다.", "fix_en": "Add a distinct red circle marking to the interior surfaces of the wheelhouse." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Open Suriyoung's eyelids halfway.\n- Remove the duplicated small photograph from Suriyoung's right hand.\n- Relax Suriyoung's right fingers completely so they rest limply without pinching.\n- Add a distinct red circle marking to the interior surfaces of the wheelhouse.\n\nPRESERVE EVERYTHING ELSE EXACTLY as 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+콘티+엔티티" }, "S30sh1": { "input_fingerprint": "10ea426bd92efdad", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): sunset, dense fog bank.\n\nSHOT TEXT (authoritative, Korean): 해질 무렵 황금빛 노을이 바다 표면에 길게 비친 전경\n\nLOCATION (lock): Open sea around a small fishing boat, with a dense offshore fog bank forming a visual boundary across the 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\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): The boat continues onward with the unconscious Su-ri-young aboard, along with her backpack, maps, old photograph, and smartphone.\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": 4, "C": 3 }, "selected": "A", "ranking": [ "A", "B", "C" ], "verdicts": [ { "label": "A", "score": 8, "verdict_ko": "어선 위라는 위치 설정과 짙은 안개, 노을의 묘사를 정확히 반영했으며 참조 이미지의 구도를 피한 점이 훌륭합니다." }, { "label": "B", "score": 4, "verdict_ko": "안개와 노을은 표현되었으나, 금지된 참조 이미지의 카메라 구도(갯벌)를 그대로 복사했고 어선이 누락되었습니다." }, { "label": "C", "score": 3, "verdict_ko": "프롬프트에 명시된 짙은 안개와 어선이 전혀 보이지 않으며, 참조 이미지의 구도를 그대로 답습했습니다." } ], "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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L18B02.png" } ], "critique": { "issues": [ { "issue_ko": "보이는 배의 갑판 위에 명시된 소지품(배낭, 지도, 오래된 사진, 스마트폰)이 전혀 보이지 않습니다.", "fix_en": "Add the backpack, maps, old photograph, and smartphone resting on the visible deck of the boat." } ] }, "fix_prompt": "Edit this photograph. Apply ONLY the corrections listed below,\neach in place, changing nothing else:\n\nCORRECTIONS:\n- Add the backpack, maps, old photograph, and smartphone resting on the visible deck 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": "플레이트만 (배경 전용)" }, "S30sh3": { "input_fingerprint": "a2bf0684c379a0cb", "prompt": "Create ONE FINAL photorealistic live-action film still of the moment below — contemporary South Korea, 2026; all people are Korean unless stated. TIME OF DAY (lock): sunset, dense fog bank.\n\nSHOT TEXT (authoritative, Korean): 어선의 뱃머리가 짙은 안개벽에 절반쯤 가려진 측면 구도\n\nLOCATION (lock): Open sea around a small fishing boat, with a dense offshore fog bank forming a visual boundary across the 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\nCARRIED STATE (persist exactly — must match the neighbouring shots of this scene): In-woo's boat continues into the fog carrying the unconscious Su-ri-young and her belongings; the wheelhouse's numerous circles remain unchanged.\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": "지시된 측면 구도와 안개에 가려진 뱃머리를 정확히 연출했으며, 금지된 레퍼런스 구도 복사를 피하면서 선박 디테일을 유지했습니다." }, { "label": "C", "score": 5, "verdict_ko": "측면 구도는 따랐으나 선박의 방향이 반대이며, 레퍼런스의 우현 디테일이 좌현에 어색하게 적용되었습니다." }, { "label": "A", "score": 4, "verdict_ko": "선박의 형태는 잘 유지했으나 측면 구도가 아니며, 레퍼런스의 카메라 구도를 그대로 모방하여 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/8ad94c03-5b33-4d51-83d0-a6e4106e9c02/episodes/e666c62b-be34-4abd-9c4d-15e4eb753eba/images/background_chain/L18B01.png" } ], "critique": { "issues": [ { "issue_ko": "레이더 돔 표면에 문자가 생성되어 있습니다.", "fix_en": "Remove all text from the radar dome." }, { "issue_ko": "조타실 창문 안쪽에 사람의 실루엣이 보입니다.", "fix_en": "Remove the human figure from inside the wheelhouse to leave it empty." } ] }, "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 radar dome.\n- Remove the human figure from inside the wheelhouse to leave it 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": "플레이트만 (배경 전용)" } }