{
 "S1sh1": {
  "input_fingerprint": "3c01ac475b05836a",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.51,
     "y": 0.56
    },
    "origin": {
     "x": 0.51,
     "y": 0.61
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "frightened running man",
     "x": 0.51,
     "y": 0.56
    }
   ],
   "rationale_ko": "남성 피해자를 A1의 빽빽한 자연 수풀 중심부에 배치했다. 카메라는 그의 진행 방향 앞쪽에서 가까운 거리를 유지하며 뒤로 이동하고, 얼굴을 정면으로 바라보도록 동일 축상에 두어 나뭇가지 사이의 타이트한 추적 구도를 만든다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "다이어그램, 마커 또는 기타 계획 기호가 보이지 않습니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.51, y=0.61) and looks toward (x=0.51, y=0.56); subjects outside the camera's horizontal field of view are out of frame\n- frightened running man stands at (x=0.51, y=0.56)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Track backward at the victim's eye level, holding his frightened face frontally in the center while nearby branches intermittently cross the foreground. Keep the framing tight on his head and shoulders as his body drives forward through the brush, preserving the frontal-pursuit axis without revealing what follows him.\n\nTHE LOCATION: 금월산 숲속\n\nSHOT TEXT (authoritative, Korean): 어두컴컴한 숲속 수풀 사이, 잔뜩 겁에 질린 표정의 금월산 남성 피해자의 얼굴이 나뭇가지 밖으로 내밀어진 정면 구도.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S1sh4": {
  "input_fingerprint": "bcae2bfbdf86f211",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.52,
     "y": 0.59
    },
    "origin": {
     "x": 0.405,
     "y": 0.705
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "airborne black silhouette with outstretched arms",
     "x": 0.485,
     "y": 0.555
    },
    {
     "slot": "E2",
     "subject_en": "fallen injured man",
     "x": 0.555,
     "y": 0.625
    }
   ],
   "rationale_ko": "A1의 빽빽한 수풀과 A2의 입목 사이, A3 woodland ground 위에 공격 순간을 배치했다. 쓰러진 피해자는 공격축의 남동쪽에 두고 검은 인영은 양팔을 뻗어 북서쪽에서 피해자 쪽으로 공중 이동하도록 놓았다. 카메라는 두 인물의 이동축 남서편 낮은 위치에서 북동쪽을 바라보게 해, 피해자가 화면 하단 우측에 놓이고 인영이 상단 좌측에서 가로지르는 측면 구도가 되도록 했다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "도면 기호, 텍스트 또는 평면도 흔적이 없는 정상적인 스케치입니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.41, y=0.70) and looks toward (x=0.52, y=0.59); subjects outside the camera's horizontal field of view are out of frame\n- airborne black silhouette with outstretched arms stands at (x=0.48, y=0.56)\n- fallen injured man stands at (x=0.56, y=0.62)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Hold a static, ground-level lateral view aimed across the fallen victim as the 검은 인영 crosses the upper portion of the frame with both arms extended. Place the victim along the lower-right plane and let the airborne silhouette travel from upper left toward him, making subject position the single decisive change within the established low attack axis.\n\nTHE LOCATION: 금월산 숲속\n\nSHOT TEXT (authoritative, Korean): 바닥에 쓰러진 금월산 남성 피해자를 덮치기 위해 양팔을 뻗은 채 공중을 가로지르는 검은 인영의 실루엣.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S2sh3": {
  "input_fingerprint": "da67b11e07a2035d",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.521,
     "y": 0.455
    },
    "origin": {
     "x": 0.389,
     "y": 0.654
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "motionless female victim",
     "x": 0.481,
     "y": 0.522
    },
    {
     "slot": "E2",
     "subject_en": "blood-mouthed Korean man",
     "x": 0.528,
     "y": 0.447
    }
   ],
   "rationale_ko": "골목 가장자리 붕괴 구역에서 여성을 전경 쪽 통로 바닥에 쓰러진 채 배치하고, 남성은 그녀 뒤이자 화면 우상단이 되는 전봇대 인근에 정지시켰다. 카메라는 통로의 좌하단에서 무릎 높이로 두 사람을 향해 고정해 여성의 몸을 전경에 두면서 고개를 든 남성을 후방 중경에 함께 담는다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "기호나 좌표 등 마커 유출 없이 정상적인 투시도로 그려졌습니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.39, y=0.65) and looks toward (x=0.52, y=0.46); subjects outside the camera's horizontal field of view are out of frame\n- motionless female victim stands at (x=0.48, y=0.52)\n- blood-mouthed Korean man stands at (x=0.53, y=0.45)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): From knee height, hold a static view across the motionless 효장동 여성 피해자 in the foreground and tilt slightly upward to retain 효장동 한국인 남성 (포식자 활성 상태) behind her. Her body stretches across the lower frame while his lifted head occupies the upper-right midground, preserving their aftermath relationship in one bleak tableau.\n\nTHE LOCATION: 도시 뒷골목\n\nSHOT TEXT (authoritative, Korean): 효장동 여성 피해자의 몸이 바닥에 쓰러진 채 멈춰 있고, 입가에 피를 묻힌 한국인 남성이 고개를 든 정지 상태.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S3sh2": {
  "input_fingerprint": "9d728f45c4d349d4",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.51,
     "y": 0.56
    },
    "origin": {
     "x": 0.475,
     "y": 0.595
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "police investigator holding a radio",
     "x": 0.51,
     "y": 0.56
    }
   ],
   "rationale_ko": "E1은 A1의 빽빽한 자연림 덤불과 A2의 수목 사이, A3의 연속된 woodland ground 위에 있는 비교적 열린 지점에 멈춰 세웠다. 카메라는 폴리스 라인 바깥쪽에서 짧게 전진해 도달한 가까운 위치로 두고, 인물을 비스듬히 바라보는 방향을 설정해 무전기를 든 상체의 안정적인 3/4 구도가 되도록 했다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "도해용 기호나 불필요한 기호가 포함되지 않은 장면입니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.47, y=0.59) and looks toward (x=0.51, y=0.56); subjects outside the camera's horizontal field of view are out of frame\n- police investigator holding a radio stands at (x=0.51, y=0.56)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Continue the dolly inward from outside the police line at 혜수's chest-to-eye level, ending on her upper body in a steady three-quarter composition with the radio held near the lower center. The sole emphasis change is camera distance, compressing the wider activity behind her into subdued broadcast-footage context while her fixed expression remains sharp.\n\nTHE LOCATION: 금월산 숲속\n\nSHOT TEXT (authoritative, Korean): 무전기를 손에 쥔 채 폴리스 라인 밖에서 굳은 표정으로 멈춰 서 있는 혜수의 상체.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S6sh1": {
  "input_fingerprint": "e0ea4c018e3cb6ea",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.41,
     "y": 0.52
    },
    "origin": {
     "x": 0.49,
     "y": 0.52
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "young woman riding a bicycle",
     "x": 0.41,
     "y": 0.52
    }
   ],
   "rationale_ko": "E1은 해안 수역(R2)과 나란히 이어지는 해안 도로 주행로(R1) 중앙에 두고 북쪽으로 주행하도록 설정했다. 카메라는 도로의 바다 쪽에서 같은 속도로 평행 이동하며 서쪽을 바라봐, 북쪽 진행이 화면상 좌→우가 되는 측면 전신 구도를 만든다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "다이어그램 기호나 텍스트 없이 스케치가 장면으로 잘 구현됨."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.49, y=0.52) and looks toward (x=0.41, y=0.52); subjects outside the camera's horizontal field of view are out of frame\n- young woman riding a bicycle stands at (x=0.41, y=0.52)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Track parallel at hip height in a wide side profile, matching 수리영's speed so her full body and bicycle remain stable across the middle of the frame while the road moves behind her. Let her travel left to right, capturing one pedal driven fully downward and her hair trailing opposite the motion without cropping either wheel.\n\nTHE LOCATION: 해안가 도로\n\nSHOT TEXT (authoritative, Korean): 맑은 낮 해안가 도로, 이어폰을 낀 수리영이 밝은 표정으로 자전거 안장에 앉아 한쪽 페달을 아래로 힘차게 밟고 있으며 주행 반대 방향으로 머리카락이 흩날리는 mid-action 측면 전신.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S9sh1": {
  "input_fingerprint": "f792c1299e5f3c2c",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.43,
     "y": 0.55
    },
    "origin": {
     "x": 0.35,
     "y": 0.55
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "young woman riding bicycle",
     "x": 0.43,
     "y": 0.55
    }
   ],
   "rationale_ko": "자전거를 탄 인물을 해안도로 주행로(R1) 중앙부에 배치했다. 카메라는 인물의 내륙 쪽에서 가까운 간격으로 나란히 이동하며 바다(R2) 방향을 바라보게 하여, 밤바다가 배경에 놓이는 토르소 높이의 중간 측면 프로필과 페달을 강하게 밟는 순간을 포착한다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "제시된 스케치에는 평면도 마커, 텍스트, 다이어그램 기호 등 정보 유출이 없습니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.35, y=0.55) and looks toward (x=0.43, y=0.55); subjects outside the camera's horizontal field of view are out of frame\n- young woman riding bicycle stands at (x=0.43, y=0.55)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Travel beside 수리영 at torso height in a medium side profile, matching the bicycle's speed so her body remains steady while the coastal road slips behind her. Catch the exact stroke with one knee folded and the opposite leg forcing its pedal downward; her profile sits forward in the frame with trailing hair and clothing edges retained behind her.\n\nTHE LOCATION: 해안가 도로\n\nSHOT TEXT (authoritative, Korean): 어두운 밤바다를 배경으로, 수리영이 굳은 표정으로 한쪽 무릎을 굽혀 자전거 페달을 힘껏 밟아 내리는 중간 동작, 옷자락과 머리카락이 뒤로 흩날리는 측면.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S13sh1": {
  "input_fingerprint": "dcffa20ba071040b",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.637,
     "y": 0.688
    },
    "origin": {
     "x": 0.57,
     "y": 0.79
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "terrified young woman running down stairs",
     "x": 0.637,
     "y": 0.688
    }
   ],
   "rationale_ko": "인물은 외부 계단(S1) 중간에서 지상 마당(Y1) 방향으로 뛰어 내려오며 한 발이 다음 계단 위에 뜬 순간이다. 카메라는 계단보다 몇 미터 아래쪽이자 마당 쪽의 평행 동선에 두고 인물을 올려다보게 하여, 전신과 인물 뒤로 이어지는 계단을 함께 담는다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "마커 기호 없이 정상 렌더링됨."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.57, y=0.79) and looks toward (x=0.64, y=0.69); subjects outside the camera's horizontal field of view are out of frame\n- terrified young woman running down stairs stands at (x=0.64, y=0.69)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): From waist height beside the exterior stairs, track downward on a parallel path several meters ahead of 수리영, holding her full body near center as she descends toward the lens. Freeze the mid-track composition with one foot suspended above the next step, her body pitched forward while the stair run remains visible beneath and behind her.\n\nTHE LOCATION: 다세대 빌라 마당 및 옥탑방 외부 계단\n\nSHOT TEXT (authoritative, Korean): 공포에 질린 얼굴의 수리영이 옥탑방 외부 계단을 뛰어내려오는 도중, 한 발이 허공에 뜬 mid-action 전신.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S13sh3": {
  "input_fingerprint": "3e58f48c6f9ee88f",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.555,
     "y": 0.738
    },
    "origin": {
     "x": 0.57,
     "y": 0.79
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "middle-aged female police officer",
     "x": 0.54,
     "y": 0.74
    },
    {
     "slot": "E2",
     "subject_en": "dazed young woman",
     "x": 0.57,
     "y": 0.735
    }
   ],
   "rationale_ko": "외부 계단(S1) 하단과 지상 마당(Y1)이 만나는 지점에 두 사람을 밀착 배치했다. 혜수는 골목과 출입문(Y2)에서 달려온 방향인 왼쪽, 수리영은 계단에서 내려온 방향인 오른쪽에 두어 서로 마주 보며 어깨를 붙잡는 순간을 구성했다. 카메라는 두 사람 남쪽의 가까운 3/4 측면에 두고 맞닿은 어깨 중앙을 바라보게 해 타이트한 상체 투샷과 좁은 외부 배경을 만든다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "도면 기호나 문자 같은 유출이 없습니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.57, y=0.79) and looks toward (x=0.56, y=0.74); subjects outside the camera's horizontal field of view are out of frame\n- middle-aged female police officer stands at (x=0.54, y=0.74)\n- dazed young woman stands at (x=0.57, y=0.73)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): At chest height in a three-quarter side position, finish the dolly-in as a tight upper-body two-shot, placing 혜수 on the left and 수리영 on the right with their joined shoulders at the center. 혜수's gripping hands form the lower visual anchor while 수리영's lifted gaze carries the emotional focus, leaving only a narrow strip of the exterior setting behind them.\n\nTHE LOCATION: 다세대 빌라 마당 및 옥탑방 외부 계단\n\nSHOT TEXT (authoritative, Korean): 혜수가 초점을 잃고 서 있는 수리영의 어깨를 쥔 채 마주 보는 상체 구도.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S15sh1": {
  "input_fingerprint": "22fea651e9451415",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.515,
     "y": 0.674
    },
    "origin": {
     "x": 0.39,
     "y": 0.82
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "seated young woman facing the sea",
     "x": 0.515,
     "y": 0.674
    }
   ],
   "rationale_ko": "인물은 A1 해안 버스정류장의 벤치 중앙에 앉혀 바다(B2) 쪽으로 고개를 향하게 했다. 카메라는 정류장보다 내륙 쪽의 접근 공간(A2) 방향에 수 미터 떨어져 두고 인물을 바라보게 해, 인물이 화면 중앙에 작게 놓이고 뒤로 해안선(B1)과 밤바다가 펼쳐지는 정적인 와이드 구도를 만든다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "기호, 텍스트, 평면도 등의 유출 요소가 발견되지 않았습니다. 지시된 카메라 앵글을 잘 따르는 스케치입니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.39, y=0.82) and looks toward (x=0.52, y=0.67); subjects outside the camera's horizontal field of view are out of frame\n- seated young woman facing the sea stands at (x=0.52, y=0.67)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): From standing eye level several meters inland, hold a static wide view facing 수리영 on the bench with the dark sea extending behind her. Keep her seated figure near the center but small within the environment, her head turned toward the water so the opening composition feels watched rather than intimate.\n\nTHE LOCATION: 해안가 버스정류장 및 인접 도로\n\nSHOT TEXT (authoritative, Korean): 어두운 밤바다 앞 버스정류장 벤치에 수리영이 멍한 표정으로 앉아 있는 모습.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S15sh4": {
  "input_fingerprint": "4172507694af1c85",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.33,
     "y": 0.52
    },
    "origin": {
     "x": 0.38,
     "y": 0.6
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "running young woman",
     "x": 0.33,
     "y": 0.52
    }
   ],
   "rationale_ko": "달리는 인물을 해안 버스정류장 인근의 육상 접근로(A2) 위에 배치하고, 정류장(A1)에서 멀어지는 좌상향 도로 축을 따라 미드 스트라이드로 달리게 했다. 카메라는 같은 도로 축상에서 수 미터 뒤인 우하측에 두어 허리 높이의 추적 촬영으로 뒷모습 전신을 중앙에 유지하도록 설정했다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "마커 기호나 평면도 유출이 없는 투시도입니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.38, y=0.60) and looks toward (x=0.33, y=0.52); subjects outside the camera's horizontal field of view are out of frame\n- running young woman stands at (x=0.33, y=0.52)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): At waist height several meters behind 수리영, track directly along her running axis on the coastal road and hold the rear full-body frame as one foot lifts in mid-stride. Her figure stays near center while the road converges ahead, maintaining the pursuit momentum without overtaking her.\n\nTHE LOCATION: 해안가 버스정류장 및 인접 도로\n\nSHOT TEXT (authoritative, Korean): 수리영이 해안가 도로를 따라 한 발이 허공에 떠 있는 미드 스트라이드(mid-stride) 자세로 달리는 뒷모습.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S23sh2": {
  "input_fingerprint": "ab48b321c9bf03db",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.418,
     "y": 0.52
    },
    "origin": {
     "x": 0.335,
     "y": 0.52
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "young woman riding bicycle",
     "x": 0.418,
     "y": 0.52
    }
   ],
   "rationale_ko": "해안 도로 여행 통로(R1) 중앙부에 자전거를 타는 여성을 배치했다. 카메라는 같은 도로 안에서 피사체의 내륙 쪽 측면에 두고 수직으로 바라보게 하여, 도로를 따라 나란히 이동하며 상체와 핸들을 담는 중간 측면 구도를 형성한다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "기호나 마커 유출 없음."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.34, y=0.52) and looks toward (x=0.42, y=0.52); subjects outside the camera's horizontal field of view are out of frame\n- young woman riding bicycle stands at (x=0.42, y=0.52)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Track parallel to 수리영 at upper-torso height from her side, matching the bicycle’s speed and holding a medium profile composition. Her forward-pitched torso occupies the center, both hands and the handlebars remain across the lower third, and her rain-swept hair and clothing trail toward the rear edge of the frame.\n\nTHE LOCATION: 해안가 도로\n\nSHOT TEXT (authoritative, Korean): 쏟아지는 빗속에서 자전거 핸들을 꽉 쥐고 상체를 앞으로 숙인 수리영의 측면, 속도에 의해 뒤로 흩날리는 젖은 머리카락과 옷자락.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S23sh3": {
  "input_fingerprint": "dcb6bbc4434ef04a",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.42,
     "y": 0.5
    },
    "origin": {
     "x": 0.465,
     "y": 0.535
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "young woman riding a bicycle in rain",
     "x": 0.42,
     "y": 0.5
    }
   ],
   "rationale_ko": "젖은 채 자전거를 타는 여성을 해안 도로 주행로(R1) 중앙부에 배치했다. 카메라는 진행 방향보다 조금 앞선 도로 안쪽에서 인물의 얼굴을 되돌아보도록 두어, 평행 이동을 유지하면서 눈과 젖은 앞머리에 집중하는 타이트한 측전면 클로즈업이 되게 했다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "기호나 텍스트 같은 마커 유출 없이 깨끗한 스케치입니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.47, y=0.54) and looks toward (x=0.42, y=0.50); subjects outside the camera's horizontal field of view are out of frame\n- young woman riding a bicycle in rain stands at (x=0.42, y=0.50)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Continue the same parallel track while moving slightly ahead and inward to an eye-level, tight side-front angle. Raise the framing just enough to hold 수리영’s eyes near center, with wet bangs crossing the upper face and only fragments of her cheek, shoulder, and receding background remaining visible.\n\nTHE LOCATION: 해안가 도로\n\nSHOT TEXT (authoritative, Korean): 수리영의 젖은 앞머리 사이로 결연한 눈빛이 번뜩이는 얼굴 클로즈업.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S25sh1": {
  "input_fingerprint": "ae8252a82d79f431",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.61,
     "y": 0.2
    },
    "origin": {
     "x": 0.34,
     "y": 0.45
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "crouched seated young woman",
     "x": 0.4,
     "y": 0.38
    }
   ],
   "rationale_ko": "인물을 R1 갯바위의 항구 입구 쪽 가장자리에 웅크려 앉히고 H3 방향의 안개 낀 바다를 바라보게 했다. 카메라는 갯바위 위에서 인물보다 수 미터 뒤쪽에 두어 뒷모습을 전경 하단에 담고, 시선축 너머로 H2의 빈 항구 입구와 넓은 바다 공간이 펼쳐지도록 배치했다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "기획 데이터 유출 기호나 마커가 발견되지 않았습니다. 자연스러운 원근감이 표현된 스케치입니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.34, y=0.45) and looks toward (x=0.61, y=0.20); subjects outside the camera's horizontal field of view are out of frame\n- crouched seated young woman stands at (x=0.40, y=0.38)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): From standing eye level several meters behind 수리영, hold a static wide frame with her crouched seated figure low in the foreground and the empty harbor opening across the background. Her back establishes a strong central axis into the fog-concealed sea, leaving broad negative space around the solitary figure before the camera begins its later descent.\n\nTHE LOCATION: 안개 낀 포구와 선착장\n\nSHOT TEXT (authoritative, Korean): 자욱한 안개가 깔린 텅 빈 포구, 수리영이 갯바위에 웅크려 앉아 바다를 응시하는 뒷모습.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S25sh2": {
  "input_fingerprint": "8d835c06a0996daa",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.321,
     "y": 0.422
    },
    "origin": {
     "x": 0.303,
     "y": 0.451
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "seated tired young woman",
     "x": 0.307,
     "y": 0.443
    },
    {
     "slot": "E2",
     "subject_en": "mobile phone resting on bent knees",
     "x": 0.321,
     "y": 0.422
    }
   ],
   "rationale_ko": "젊은 여성은 항구 입구와 외해를 바라볼 수 있는 R1 해안 갯바위 대기 지점에 앉혀 두었다. 휴대전화는 바다 쪽으로 굽힌 무릎 중앙에 놓았으며, 카메라는 무릎의 바로 뒤쪽 상공에서 휴대전화를 향해 내려다보는 급한 하이앵글 클로즈업이 되도록 매우 가깝게 배치했다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "도면 마커나 다이어그램 기호 없이 자연스럽게 묘사됨."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.30, y=0.45) and looks toward (x=0.32, y=0.42); subjects outside the camera's horizontal field of view are out of frame\n- seated tired young woman stands at (x=0.31, y=0.44)\n- mobile phone resting on bent knees stands at (x=0.32, y=0.42)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Complete the descending dolly directly above and slightly behind 수리영’s knees, settling into a steep high-angle close view. The phone lies centrally over her bent knees and occupies the lower central portion of the frame, large enough for 혜수’s repeated missed-call notifications to read while the knees retain a clear human scale reference.\n\nTHE LOCATION: 안개 낀 포구와 선착장\n\nSHOT TEXT (authoritative, Korean): 수리영의 무릎 위에서 휴대전화 화면에 혜수의 부재중 전화 알림이 떠 있는 클로즈업.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S25sh6": {
  "input_fingerprint": "ea419bc0207754ee",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.565,
     "y": 0.621
    },
    "origin": {
     "x": 0.603,
     "y": 0.659
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "young man holding a mooring rope",
     "x": 0.566,
     "y": 0.623
    },
    {
     "slot": "E2",
     "subject_en": "young woman standing behind him",
     "x": 0.594,
     "y": 0.651
    }
   ],
   "rationale_ko": "인우는 H1 항구 착선 가장자리에서 계류 밧줄을 쥔 채 물가 쪽에 배치하고, 수리영은 그의 반쯤 돌아간 시선 방향과 카메라 사이의 육지 쪽 가까운 위치에 두었다. 카메라는 수리영의 어깨 너머로 인우의 무표정한 얼굴을 근접 포착하도록 인우 옆의 낮은 각도에 두며, 화면 가장자리에는 착선 가장자리와 이어지는 밧줄이 남을 수 있게 했다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "제시된 스케치에는 평면도 마커, 화살표, 문자 등 스테이징 데이터 유출 기호가 없습니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.60, y=0.66) and looks toward (x=0.56, y=0.62); subjects outside the camera's horizontal field of view are out of frame\n- young man holding a mooring rope stands at (x=0.57, y=0.62)\n- young woman standing behind him stands at (x=0.59, y=0.65)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Finish the tracking arc at face level beside 인우, turning his half-rotated face into the close primary subject while retaining the mooring rope along one frame edge. 수리영 remains as a restrained over-shoulder presence in the near foreground, preserving the direction of his glance without competing with his unreadable expression.\n\nTHE LOCATION: 안개 낀 포구와 선착장\n\nSHOT TEXT (authoritative, Korean): 인우가 밧줄을 쥔 채 고개를 반쯤 돌려 수리영을 힐끗 쳐다보고 있는 무표정한 얼굴.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S27sh4": {
  "input_fingerprint": "d448c812e9360901",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.64,
     "y": 0.61
    },
    "origin": {
     "x": 0.49,
     "y": 0.69
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "seated young woman pointing at a photograph",
     "x": 0.53,
     "y": 0.65
    },
    {
     "slot": "E2",
     "subject_en": "unsettled man facing the woman",
     "x": 0.65,
     "y": 0.61
    }
   ],
   "rationale_ko": "두 인물을 D1 외부 갑판 중앙에 마주 보게 배치했다. 카메라는 앉아 있는 여성의 좌후방 가까이에서 남성을 향해 비스듬히 촬영하여, 여성의 상체와 가리키는 손이 좌측 전경에 놓이고 사진은 두 사람 사이의 하단 중앙, 남성은 우측 중경에 보이도록 했다. W2 조타실 입구로 향해 달려가기 전의 대치 순간이므로 입구 앞 동선은 비워 두었다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "마커나 기호, 텍스트 등 기획 데이터 유출 흔적이 발견되지 않았습니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.49, y=0.69) and looks toward (x=0.64, y=0.61); subjects outside the camera's horizontal field of view are out of frame\n- seated young woman pointing at a photograph stands at (x=0.53, y=0.65)\n- unsettled man facing the woman stands at (x=0.65, y=0.61)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Dolly inward from beside the seated 수리영 at chest height, angling across the photograph toward 인우 while preserving both ends of her accusatory eyeline. 수리영’s upper body and pointing hand occupy the left foreground, the old photograph rests near the lower center, and 인우’s unsettled face remains across from her in the right midground.\n\nTHE LOCATION: 소형 어선 갑판\n\nSHOT TEXT (authoritative, Korean): 수리영이 낡은 흑백 사진을 손가락으로 짚은 채 인우의 눈을 뚫어지게 응시하는 상체.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 },
 "S29sh3": {
  "input_fingerprint": "0b3afbb2cb478fcd",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.62,
     "y": 0.48
    },
    "origin": {
     "x": 0.62,
     "y": 0.72
    }
   },
   "entity_placements": [
    {
     "slot": "E1",
     "subject_en": "moving boat entering fog",
     "x": 0.62,
     "y": 0.48
    }
   ],
   "rationale_ko": "A1 오픈 시의 중앙부에서 배가 지도상 오른쪽으로 전진하도록 배치했다. 카메라는 배의 전방 절반 옆쪽 해상에 거리를 두고 평행 이동하며 측면 와이드 숏을 얻도록 배의 남쪽에서 뱃머리 방향을 바라본다."
  },
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "도해 기호나 텍스트 누출이 발견되지 않음"
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached TOP-DOWN SITE PLAN, seen from the camera position\ngiven in the STAGING GEOMETRY data below.\n\nJurisdiction of sources — follow all three, they do not conflict:\n- The attached SITE PLAN is the SOT for what exists on the site and\n  how its zones and landmarks are arranged relative to each other.\n- The STAGING GEOMETRY data below is the SOT for XY POSITIONS and\n  horizontal heading ONLY: where each subject stands on that plan,\n  where the camera stands, and which way it looks. Distances,\n  directions and relative depths between subjects come from these\n  coordinates. Subjects outside the camera's horizontal field of\n  view are out of frame.\n- The CAMERA DIRECTION text below (when present) is the SOT for the\n  VERTICAL angle, elevation, tilt and framing (eye-level, high angle,\n  overhead crane, low angle, ...). If it asks for a high or overhead\n  angle, draw that angle — the site plan being top-down does NOT mean\n  the frame must be eye-level.\n\nSTYLE — LIGHT STORYBOARD FRAME: thin uniform dark outlines on plain\npaper. No shading, no hatching, no texture, no facial features beyond\na minimal head shape, no costume detail. Human figures are correctly\nproportioned full-body outlines in their canonical poses. Environment\nis reduced to a few placement lines (ground line, key landmark\nsilhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow are normalized map coordinates on the attached site plan\n(x runs left→right 0..1, y runs top→bottom 0..1). Use them solely to\nplace the camera and subjects. Do NOT draw any of this data in the\nimage — no numbers, no coordinates, no labels, no arrows, no cones,\nno dots, no diagram symbols of any kind.\n- the camera stands at (x=0.62, y=0.72) and looks toward (x=0.62, y=0.48); subjects outside the camera's horizontal field of view are out of frame\n- moving boat entering fog stands at (x=0.62, y=0.48)\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): At sea level beside the forward half of the boat, the tracking camera moves parallel to its course in a wide side view as the bow advances from screen left toward the dense fog on screen right. The bow and its wake remain below forty percent of the composition, leaving enough sea and fog visible to show the boat's forward half becoming obscured at the end of the move.\n\nTHE LOCATION: 서해 바다\n\nSHOT TEXT (authoritative, Korean): 배의 앞머리 절반이 짙은 안개 속에 파묻힌 채, 뱃머리에서 물살을 일으키며 안개 속으로 진입하는 순간의 측면 컷.\n\nCRITICAL — THE SITE PLAN IS A STAGING REFERENCE, NOT CONTENT: do NOT\ndraw the plan itself, and do NOT draw any diagram elements in the\nframe — no circles, dots, wedges, cones, arrows, callouts, codes or\nany other overlay symbols. The output is a clean storyboard drawing\nof the scene as seen by the camera only.\n\nNo text, no letters, no numbers, no captions anywhere in the image."
 }
}