{
 "S7sh9": {
  "input_fingerprint": "271b84aa2a0a53bf",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.66,
     "y": 0.54
    },
    "look_target_anchor_en": "In the central-right parking aisle below the station-side pedestrian entrance.",
    "origin": {
     "x": 0.31,
     "y": 0.72
    },
    "origin_anchor_en": "Just inside the lower-left boundary of the police station parking area beside the curved vehicle approach.",
    "view_left": {
     "x": 0.42,
     "y": 0.45
    },
    "view_left_anchor_en": "Along the building-facing edge of the parking area, well left of the station-side pedestrian entrance.",
    "view_right": {
     "x": 0.9,
     "y": 0.64
    },
    "view_right_anchor_en": "Along the right boundary of the police station parking area below the outermost parking bay."
   },
   "entity_placements": [
    {
     "anchor_en": "In the right half of the police station parking area, several bays below the building-facing edge.",
     "camera_facing_evidence": "등을 돌려 저만치 멀어진 강민숙(홍은미)의 뒷모습",
     "camera_facing_relation": "away_from_camera",
     "facing": {
      "x": 0.86,
      "y": 0.45
     },
     "facing_anchor_en": "At the building-facing edge of the parking area to the right of the station-side pedestrian entrance.",
     "slot": "E1",
     "subject_en": "receding middle-aged woman",
     "x": 0.75,
     "y": 0.52
    },
    {
     "anchor_en": "In the lower-left part of the police station parking area near the curved vehicle-approach boundary.",
     "camera_facing_evidence": "behind and slightly to one side of 정인우",
     "camera_facing_relation": "away_from_camera",
     "facing": {
      "x": 0.76,
      "y": 0.51
     },
     "facing_anchor_en": "In the right half of the parking area below the building-facing parking bays.",
     "slot": "E2",
     "subject_en": "motionless man holding a money envelope",
     "x": 0.4,
     "y": 0.65
    }
   ],
   "rationale_ko": "카메라를 인우의 뒤쪽이자 약간 옆인 주차장 좌하단에 두고, 인우를 좌측 전경에 전신으로 배치했다. 민숙은 우측 후경에서 건물 쪽으로 멀어지게 하여 두 사람 사이의 긴 거리가 넓은 화면의 중심 긴장이 되도록 구성했다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle Just inside the lower-left boundary of the police station parking area beside the curved vehicle approach. (coarse estimate x=0.31, y=0.72); its sight line ends at LOOK TARGET: In the central-right parking aisle below the station-side pedestrian entrance. (coarse estimate x=0.66, y=0.54); the frame's LEFT edge cuts through Along the building-facing edge of the parking area, well left of the station-side pedestrian entrance. (coarse estimate x=0.42, y=0.45) and its RIGHT edge through Along the right boundary of the police station parking area below the outermost parking bay. (coarse estimate x=0.90, y=0.64)\n- E1 (receding middle-aged woman): circle In the right half of the police station parking area, several bays below the building-facing edge. (coarse estimate x=0.75, y=0.52), facing At the building-facing edge of the parking area to the right of the station-side pedestrian entrance. (coarse estimate x=0.86, y=0.45)\n- E2 (motionless man holding a money envelope): circle In the lower-left part of the police station parking area near the curved vehicle-approach boundary. (coarse estimate x=0.40, y=0.65), facing In the right half of the parking area below the building-facing parking bays. (coarse estimate x=0.76, y=0.51)",
  "marker_attempts": 1,
  "marker_map_sha256": "07ac7596233af236a3616c3e7b1b37610435d7181cd47047c62b013fb64755dc",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "도면 기호나 평면도 같은 유출이 없습니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands Just inside the lower-left boundary of the police station parking area beside the curved vehicle approach. (coarse estimate x=0.31, y=0.72); its sight line ends at LOOK TARGET: In the central-right parking aisle below the station-side pedestrian entrance. (coarse estimate x=0.66, y=0.54); the frame's LEFT edge cuts through Along the building-facing edge of the parking area, well left of the station-side pedestrian entrance. (coarse estimate x=0.42, y=0.45) and its RIGHT edge through Along the right boundary of the police station parking area below the outermost parking bay. (coarse estimate x=0.90, y=0.64) — everything between those two edges is in frame, anything outside them is out of frame\n- receding middle-aged woman stands In the right half of the police station parking area, several bays below the building-facing edge. (coarse estimate x=0.75, y=0.52), facing At the building-facing edge of the parking area to the right of the station-side pedestrian entrance. (coarse estimate x=0.86, y=0.45) — from this camera the subject is near the CENTER of the frame and faces away from the camera, into the depth of the shot, so the camera sees this subject from behind\n- motionless man holding a money envelope stands In the lower-left part of the police station parking area near the curved vehicle-approach boundary. (coarse estimate x=0.40, y=0.65), facing In the right half of the parking area below the building-facing parking bays. (coarse estimate x=0.76, y=0.51) — from this camera the subject is in the LEFT part of the frame and faces away from the camera, into the depth of the shot, so the camera sees this subject from behind; this subject is turned toward the receding middle-aged woman\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): From eye level behind and slightly to one side of 정인우, complete the dolly-out into a wide composition that holds his full body in the left foreground and 강민숙(홍은미) receding in the right background. His lowered shoulders and money-envelope hand remain readable, while the distance between them becomes the dominant graphic tension of the parking area.\n\nTHE LOCATION: 경찰서 주차장\n\nSHOT TEXT (authoritative, Korean): 한 손에 돈봉투를 든 채 등을 돌려 저만치 멀어진 강민숙(홍은미)의 뒷모습을 우두커니 서서 멍하니 바라보는 정인우의 전신.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S20sh6": {
  "input_fingerprint": "3794fedec972bce3",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.5,
     "y": 0.91
    },
    "look_target_anchor_en": "At the center of the outer pier end beside the seaward terminal boundary.",
    "origin": {
     "x": 0.5,
     "y": 0.54
    },
    "origin_anchor_en": "On the centerline of the main island pier deck just seaward of the village-side approach.",
    "view_left": {
     "x": 0.39,
     "y": 0.96
    },
    "view_left_anchor_en": "At the left seaward corner of the outer pier end where the deck meets the sea.",
    "view_right": {
     "x": 0.61,
     "y": 0.96
    },
    "view_right_anchor_en": "At the right seaward corner of the outer pier end where the deck meets the sea."
   },
   "entity_placements": [
    {
     "anchor_en": "On the centerline of the main island pier deck near its village-side connection.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.5,
      "y": 0.9
     },
     "facing_anchor_en": "At the center of the outer pier end beside the seaward terminal boundary.",
     "slot": "E1",
     "subject_en": "approaching young woman",
     "x": 0.5,
     "y": 0.59
    },
    {
     "anchor_en": "At the center of the outer pier end with clear deck space around her.",
     "camera_facing_evidence": "강영란(흰 머리 여자)의 정면 전신",
     "camera_facing_relation": "toward_camera",
     "facing": {
      "x": 0.5,
      "y": 0.64
     },
     "facing_anchor_en": "On the centerline of the main island pier deck between the outer pier end and the village-side approach.",
     "slot": "E2",
     "subject_en": "smiling white-haired elderly woman",
     "x": 0.5,
     "y": 0.9
    }
   ],
   "rationale_ko": "카메라는 지후의 접근 시선과 같은 축으로 부두 중앙에서 외항 끝을 바라본다. 지후를 카메라 바로 앞의 부두 중앙에 두고, 영란은 외항 끝 중앙에 홀로 세워 두 사람 사이의 빈 공간과 부두의 원근선을 강조했다. 영란은 카메라 쪽인 육지 방향을 바라보므로 정면 전신과 렌즈를 마주하는 미소가 성립하며, 두 인물 모두 시야 쐐기 안에 배치했다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle On the centerline of the main island pier deck just seaward of the village-side approach. (coarse estimate x=0.50, y=0.54); its sight line ends at LOOK TARGET: At the center of the outer pier end beside the seaward terminal boundary. (coarse estimate x=0.50, y=0.91); the frame's LEFT edge cuts through At the left seaward corner of the outer pier end where the deck meets the sea. (coarse estimate x=0.39, y=0.96) and its RIGHT edge through At the right seaward corner of the outer pier end where the deck meets the sea. (coarse estimate x=0.61, y=0.96)\n- E1 (approaching young woman): circle On the centerline of the main island pier deck near its village-side connection. (coarse estimate x=0.50, y=0.59), facing At the center of the outer pier end beside the seaward terminal boundary. (coarse estimate x=0.50, y=0.90)\n- E2 (smiling white-haired elderly woman): circle At the center of the outer pier end with clear deck space around her. (coarse estimate x=0.50, y=0.90), facing On the centerline of the main island pier deck between the outer pier end and the village-side approach. (coarse estimate x=0.50, y=0.64)",
  "marker_attempts": 1,
  "marker_map_sha256": "b3938d644842585bd10ef7f8c4f47c6d80ede980bdb412c4fe99d7ed51661aae",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "화살표, 문자 등 도면 기호가 없는 일반 스케치입니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands On the centerline of the main island pier deck just seaward of the village-side approach. (coarse estimate x=0.50, y=0.54); its sight line ends at LOOK TARGET: At the center of the outer pier end beside the seaward terminal boundary. (coarse estimate x=0.50, y=0.91); the frame's LEFT edge cuts through At the left seaward corner of the outer pier end where the deck meets the sea. (coarse estimate x=0.39, y=0.96) and its RIGHT edge through At the right seaward corner of the outer pier end where the deck meets the sea. (coarse estimate x=0.61, y=0.96) — everything between those two edges is in frame, anything outside them is out of frame\n- approaching young woman stands On the centerline of the main island pier deck near its village-side connection. (coarse estimate x=0.50, y=0.59), facing At the center of the outer pier end beside the seaward terminal boundary. (coarse estimate x=0.50, y=0.90) — from this camera the subject is near the CENTER of the frame and faces away from the camera, into the depth of the shot, so the camera sees this subject from behind; this subject is turned toward the smiling white-haired elderly woman\n- smiling white-haired elderly woman stands At the center of the outer pier end with clear deck space around her. (coarse estimate x=0.50, y=0.90), facing On the centerline of the main island pier deck between the outer pier end and the village-side approach. (coarse estimate x=0.50, y=0.64) — from this camera the subject is near the CENTER of the frame and faces toward the camera, so the camera sees this subject from the front; this subject is turned toward the approaching young woman\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): From 김지후(강수리영)'s approaching eye line, hold a static wide subjective view with 강영란(흰 머리 여자)'s full figure centered at the end of the dock and clear space separating her from the crowd. Her direct smile meets the lens as Jihoo would see it, while the dock's receding line makes the apparently warm welcome feel ceremonially arranged.\n\nTHE LOCATION: 금월도 선착장\n\nSHOT TEXT (authoritative, Korean): 김지후(강수리영)의 시선 끝에 온화한 미소를 띠고 선 강영란(흰 머리 여자)의 정면 전신.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S60sh5": {
  "input_fingerprint": "2543a0bf1c071cd8",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.51,
     "y": 0.61
    },
    "look_target_anchor_en": "At the seaward edge of the tidal shoreline rock outcrops where the rocks meet the coastal sea.",
    "origin": {
     "x": 0.4,
     "y": 0.41
    },
    "origin_anchor_en": "On the narrow coastal walking path just seaward of the steep forestward rock passage.",
    "view_left": {
     "x": 0.76,
     "y": 0.65
    },
    "view_left_anchor_en": "At the right-hand seaward edge of the tidal shoreline rock outcrops beside the coastal sea.",
    "view_right": {
     "x": 0.25,
     "y": 0.73
    },
    "view_right_anchor_en": "In the coastal sea just beyond the left-hand tidal shoreline rock outcrops."
   },
   "entity_placements": [
    {
     "anchor_en": "Wedged among the tidal shoreline rock outcrops at their boundary with the coastal sea.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.57,
      "y": 0.72
     },
     "facing_anchor_en": "In the coastal sea just beyond the tidal shoreline rock outcrops.",
     "slot": "E1",
     "subject_en": "slack human corpse",
     "x": 0.51,
     "y": 0.61
    }
   ],
   "rationale_ko": "시신을 갯바위와 바닷물이 맞닿는 경계에 배치해 바위 틈에 걸린 상태를 구현했다. 카메라는 해안길에서 바다 쪽으로 넓게 바라보며 시신 전신, 주변 바위, 그 너머의 파도와 물보라가 함께 프레임에 들어오도록 설정했다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle On the narrow coastal walking path just seaward of the steep forestward rock passage. (coarse estimate x=0.40, y=0.41); its sight line ends at LOOK TARGET: At the seaward edge of the tidal shoreline rock outcrops where the rocks meet the coastal sea. (coarse estimate x=0.51, y=0.61); the frame's LEFT edge cuts through At the right-hand seaward edge of the tidal shoreline rock outcrops beside the coastal sea. (coarse estimate x=0.76, y=0.65) and its RIGHT edge through In the coastal sea just beyond the left-hand tidal shoreline rock outcrops. (coarse estimate x=0.25, y=0.73)\n- E1 (slack human corpse): circle Wedged among the tidal shoreline rock outcrops at their boundary with the coastal sea. (coarse estimate x=0.51, y=0.61), facing In the coastal sea just beyond the tidal shoreline rock outcrops. (coarse estimate x=0.57, y=0.72)",
  "marker_attempts": 1,
  "marker_map_sha256": "2509ee9aed223742955b2e7a096475ad7d6494ccd32e722d5203f93185168825",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "기호나 텍스트 같은 유출 요소가 발견되지 않음."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands On the narrow coastal walking path just seaward of the steep forestward rock passage. (coarse estimate x=0.40, y=0.41); its sight line ends at LOOK TARGET: At the seaward edge of the tidal shoreline rock outcrops where the rocks meet the coastal sea. (coarse estimate x=0.51, y=0.61); the frame's LEFT edge cuts through At the right-hand seaward edge of the tidal shoreline rock outcrops beside the coastal sea. (coarse estimate x=0.76, y=0.65) and its RIGHT edge through In the coastal sea just beyond the left-hand tidal shoreline rock outcrops. (coarse estimate x=0.25, y=0.73) — everything between those two edges is in frame, anything outside them is out of frame\n- slack human corpse stands Wedged among the tidal shoreline rock outcrops at their boundary with the coastal sea. (coarse estimate x=0.51, y=0.61), facing In the coastal sea just beyond the tidal shoreline rock outcrops. (coarse estimate x=0.57, y=0.72) — from this camera the subject is near the CENTER of the frame and faces away from the camera, into the depth of the shot, so the camera sees this subject from behind\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Maintain the previous eye-level pivot point and pan along 김지후(강수리영)'s seaward eyeline, settling into a wide composition that reveals 김여상's entire slack posture caught between the rocks and seawater. Place the body across the middle of the frame with rocks providing scale below and the breaking water and suspended spray visible beyond, avoiding any detached forensic close-up.\n\nTHE LOCATION: 금월도 해안 바윗길\n\nSHOT TEXT (authoritative, Korean): 거친 파도가 갯바위에 부딪혀 하얀 물보라가 공중에 멈춘 순간, 바위 틈에 기괴한 자세로 널브러져 있는 김여상의 축 늘어진 시체 전경.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S61sh15": {
  "input_fingerprint": "8cad5aa8c96ee01b",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.5,
     "y": 0.61
    },
    "look_target_anchor_en": "On the village-side half of the main pier deck near its centerline.",
    "origin": {
     "x": 0.54,
     "y": 0.74
    },
    "origin_anchor_en": "On the central main pier deck, several steps seaward of the village-side approach.",
    "view_left": {
     "x": 0.42,
     "y": 0.59
    },
    "view_left_anchor_en": "On the left half of the main pier deck near its village-side section.",
    "view_right": {
     "x": 0.62,
     "y": 0.59
    },
    "view_right_anchor_en": "Just beyond the water-contacting vessel berth edge beside the village-side section of the pier."
   },
   "entity_placements": [
    {
     "anchor_en": "On the village-side half of the main pier deck near its centerline.",
     "camera_facing_evidence": "정인우를 향해",
     "camera_facing_relation": "toward_camera",
     "facing": {
      "x": 0.56,
      "y": 0.66
     },
     "facing_anchor_en": "On the right half of the main pier deck a short distance inside the vessel berth edge.",
     "slot": "E1",
     "subject_en": "white-haired woman blocking the route",
     "x": 0.5,
     "y": 0.61
    },
    {
     "anchor_en": "On the right half of the main pier deck a short distance inside the vessel berth edge.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.5,
      "y": 0.61
     },
     "facing_anchor_en": "On the village-side half of the main pier deck near its centerline.",
     "slot": "E2",
     "subject_en": "man halted at the edge of the confrontation",
     "x": 0.56,
     "y": 0.66
    }
   ],
   "rationale_ko": "흰 머리 여자를 부두 중앙의 진행선에 세우고 남자를 선착장 가장자리 쪽에 두어 한 손으로 길을 막는 관계가 드러나게 했다. 카메라는 남자 쪽 뒤편에서 여자의 상체를 바라보며, 여자는 카메라와 남자 방향을 향하고 남자는 프레임 한쪽 가장자리에 걸리도록 구성했다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle On the central main pier deck, several steps seaward of the village-side approach. (coarse estimate x=0.54, y=0.74); its sight line ends at LOOK TARGET: On the village-side half of the main pier deck near its centerline. (coarse estimate x=0.50, y=0.61); the frame's LEFT edge cuts through On the left half of the main pier deck near its village-side section. (coarse estimate x=0.42, y=0.59) and its RIGHT edge through Just beyond the water-contacting vessel berth edge beside the village-side section of the pier. (coarse estimate x=0.62, y=0.59)\n- E1 (white-haired woman blocking the route): circle On the village-side half of the main pier deck near its centerline. (coarse estimate x=0.50, y=0.61), facing On the right half of the main pier deck a short distance inside the vessel berth edge. (coarse estimate x=0.56, y=0.66)\n- E2 (man halted at the edge of the confrontation): circle On the right half of the main pier deck a short distance inside the vessel berth edge. (coarse estimate x=0.56, y=0.66), facing On the village-side half of the main pier deck near its centerline. (coarse estimate x=0.50, y=0.61)",
  "marker_attempts": 1,
  "marker_map_sha256": "15ab86f2a90035ca5b378edac63e6d064ce6f87b54742546fd520a4cd2045aa4",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "기호나 마커 없이 정상적인 3D 뷰로 그려졌습니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands On the central main pier deck, several steps seaward of the village-side approach. (coarse estimate x=0.54, y=0.74); its sight line ends at LOOK TARGET: On the village-side half of the main pier deck near its centerline. (coarse estimate x=0.50, y=0.61); the frame's LEFT edge cuts through On the left half of the main pier deck near its village-side section. (coarse estimate x=0.42, y=0.59) and its RIGHT edge through Just beyond the water-contacting vessel berth edge beside the village-side section of the pier. (coarse estimate x=0.62, y=0.59) — everything between those two edges is in frame, anything outside them is out of frame\n- white-haired woman blocking the route stands On the village-side half of the main pier deck near its centerline. (coarse estimate x=0.50, y=0.61), facing On the right half of the main pier deck a short distance inside the vessel berth edge. (coarse estimate x=0.56, y=0.66) — from this camera the subject is near the CENTER of the frame and faces toward the RIGHT of the frame and toward the camera, so the camera sees this subject from the front; this subject is turned toward the man halted at the edge of the confrontation\n- man halted at the edge of the confrontation stands On the right half of the main pier deck a short distance inside the vessel berth edge. (coarse estimate x=0.56, y=0.66), facing On the village-side half of the main pier deck near its centerline. (coarse estimate x=0.50, y=0.61) — from this camera the subject is in the RIGHT part of the frame and faces toward the LEFT of the frame and away from the camera, into the depth of the shot, so the camera sees this subject from behind; this subject is turned toward the white-haired woman blocking the route\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): From 정인우's side of the gathering, settle the pan into a firm eye-level medium frame centered on 강영란(흰 머리 여자)'s raised palm and unyielding upper body. Her hand occupies the central foreground without exceeding natural scale, while 정인우 remains checked at one edge behind the opposing line, making the blocked route readable within a single plane.\n\nTHE LOCATION: 금월도 선착장\n\nSHOT TEXT (authoritative, Korean): 정인우를 향해 한 손을 꼿꼿이 들어 올린 채 단호하게 앞을 막아선 강영란(흰 머리 여자)의 굳건한 상체.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S65sh1": {
  "input_fingerprint": "7f43f2e7f1c216aa",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.49,
     "y": 0.67
    },
    "look_target_anchor_en": "The camera looks toward the center of the restaurant-front alley directly outside the front entrance.",
    "origin": {
     "x": 0.86,
     "y": 0.72
    },
    "origin_anchor_en": "The camera stands near the right end of the restaurant-front alley beside the village-road approach.",
    "view_left": {
     "x": 0.25,
     "y": 0.82
    },
    "view_left_anchor_en": "The left edge of the view reaches the left portion of the uphill village road below the alley.",
    "view_right": {
     "x": 0.25,
     "y": 0.55
    },
    "view_right_anchor_en": "The right edge of the view reaches the lower-left portion of the restaurant main building."
   },
   "entity_placements": [
    {
     "anchor_en": "This youth stands near the right end of the restaurant-front alley opposite the restaurant façade.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.72,
      "y": 0.62
     },
     "facing_anchor_en": "The facing point is on the right half of the restaurant front façade.",
     "slot": "E1",
     "subject_en": "watchful Korean village youth",
     "x": 0.77,
     "y": 0.71
    },
    {
     "anchor_en": "This youth stands in the right half of the restaurant-front alley several steps from the front entrance.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.62,
      "y": 0.62
     },
     "facing_anchor_en": "The facing point is on the restaurant front façade just right of the front entrance.",
     "slot": "E2",
     "subject_en": "watchful Korean village youth",
     "x": 0.66,
     "y": 0.74
    },
    {
     "anchor_en": "This youth stands near the center of the restaurant-front alley opposite the front entrance.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.53,
      "y": 0.62
     },
     "facing_anchor_en": "The facing point is on the restaurant front façade beside the front entrance.",
     "slot": "E3",
     "subject_en": "watchful Korean village youth",
     "x": 0.56,
     "y": 0.68
    },
    {
     "anchor_en": "This youth stands in the left half of the restaurant-front alley opposite the restaurant façade.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.46,
      "y": 0.62
     },
     "facing_anchor_en": "The facing point is on the restaurant front façade just left of the front entrance.",
     "slot": "E4",
     "subject_en": "watchful Korean village youth",
     "x": 0.44,
     "y": 0.73
    },
    {
     "anchor_en": "This youth stands near the left end of the restaurant-front alley beside the building frontage.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.35,
      "y": 0.62
     },
     "facing_anchor_en": "The facing point is on the left half of the restaurant front façade.",
     "slot": "E5",
     "subject_en": "watchful Korean village youth",
     "x": 0.32,
     "y": 0.68
    }
   ],
   "rationale_ko": "카메라는 골목 오른쪽 끝에서 식당 방향으로 넓게 바라보며, 가장 가까운 청년을 우측 외곽에 두고 나머지를 불규칙한 간격으로 골목 전역에 분산했다. 모든 청년의 몸 방향은 식당 전면을 향하게 하여 감시가 조직적으로 느껴지도록 했고, 인물 사이의 빈 공간을 충분히 확보했다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle The camera stands near the right end of the restaurant-front alley beside the village-road approach. (coarse estimate x=0.86, y=0.72); its sight line ends at LOOK TARGET: The camera looks toward the center of the restaurant-front alley directly outside the front entrance. (coarse estimate x=0.49, y=0.67); the frame's LEFT edge cuts through The left edge of the view reaches the left portion of the uphill village road below the alley. (coarse estimate x=0.25, y=0.82) and its RIGHT edge through The right edge of the view reaches the lower-left portion of the restaurant main building. (coarse estimate x=0.25, y=0.55)\n- E1 (watchful Korean village youth): circle This youth stands near the right end of the restaurant-front alley opposite the restaurant façade. (coarse estimate x=0.77, y=0.71), facing The facing point is on the right half of the restaurant front façade. (coarse estimate x=0.72, y=0.62)\n- E2 (watchful Korean village youth): circle This youth stands in the right half of the restaurant-front alley several steps from the front entrance. (coarse estimate x=0.66, y=0.74), facing The facing point is on the restaurant front façade just right of the front entrance. (coarse estimate x=0.62, y=0.62)\n- E3 (watchful Korean village youth): circle This youth stands near the center of the restaurant-front alley opposite the front entrance. (coarse estimate x=0.56, y=0.68), facing The facing point is on the restaurant front façade beside the front entrance. (coarse estimate x=0.53, y=0.62)\n- E4 (watchful Korean village youth): circle This youth stands in the left half of the restaurant-front alley opposite the restaurant façade. (coarse estimate x=0.44, y=0.73), facing The facing point is on the restaurant front façade just left of the front entrance. (coarse estimate x=0.46, y=0.62)\n- E5 (watchful Korean village youth): circle This youth stands near the left end of the restaurant-front alley beside the building frontage. (coarse estimate x=0.32, y=0.68), facing The facing point is on the left half of the restaurant front façade. (coarse estimate x=0.35, y=0.62)",
  "marker_attempts": 1,
  "marker_map_sha256": "781579204f12bacd5433161a48e5a7833631cbd49dafbe49586fb17ac7ea4797",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "마커나 기호가 없는 정상적인 투시도입니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands The camera stands near the right end of the restaurant-front alley beside the village-road approach. (coarse estimate x=0.86, y=0.72); its sight line ends at LOOK TARGET: The camera looks toward the center of the restaurant-front alley directly outside the front entrance. (coarse estimate x=0.49, y=0.67); the frame's LEFT edge cuts through The left edge of the view reaches the left portion of the uphill village road below the alley. (coarse estimate x=0.25, y=0.82) and its RIGHT edge through The right edge of the view reaches the lower-left portion of the restaurant main building. (coarse estimate x=0.25, y=0.55) — everything between those two edges is in frame, anything outside them is out of frame\n- watchful Korean village youth stands This youth stands near the right end of the restaurant-front alley opposite the restaurant façade. (coarse estimate x=0.77, y=0.71), facing The facing point is on the right half of the restaurant front façade. (coarse estimate x=0.72, y=0.62) — from this camera the subject is near the CENTER of the frame and faces toward the RIGHT of the frame and away from the camera, into the depth of the shot, so the camera sees this subject from behind\n- watchful Korean village youth stands This youth stands in the right half of the restaurant-front alley several steps from the front entrance. (coarse estimate x=0.66, y=0.74), facing The facing point is on the restaurant front façade just right of the front entrance. (coarse estimate x=0.62, y=0.62) — from this camera the subject is in the LEFT part of the frame and faces toward the RIGHT of the frame and away from the camera, into the depth of the shot, so the camera sees this subject from behind\n- watchful Korean village youth stands This youth stands near the center of the restaurant-front alley opposite the front entrance. (coarse estimate x=0.56, y=0.68), facing The facing point is on the restaurant front façade beside the front entrance. (coarse estimate x=0.53, y=0.62) — from this camera the subject is near the CENTER of the frame and faces toward the RIGHT of the frame and away from the camera, into the depth of the shot, so the camera sees this subject from behind\n- watchful Korean village youth stands This youth stands in the left half of the restaurant-front alley opposite the restaurant façade. (coarse estimate x=0.44, y=0.73), facing The facing point is on the restaurant front façade just left of the front entrance. (coarse estimate x=0.46, y=0.62) — from this camera the subject is near the CENTER of the frame and faces toward the RIGHT of the frame\n- watchful Korean village youth stands This youth stands near the left end of the restaurant-front alley beside the building frontage. (coarse estimate x=0.32, y=0.68), facing The facing point is on the left half of the restaurant front façade. (coarse estimate x=0.35, y=0.62) — from this camera the subject is near the CENTER of the frame and faces toward the RIGHT of the frame and toward the camera, so the camera sees this subject from the front\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): At eye level near one end of the alley, hold a static wide view down the dispersed line of Korean village youths toward the restaurant. Place the nearest youth on one outer third and the others at irregular midground intervals, all sharing the restaurant as their eyeline while negative space between them makes the surveillance feel coordinated rather than casual.\n\nTHE LOCATION: 반월 식당 앞과 골목\n\nSHOT TEXT (authoritative, Korean): 어두운 식당 밖 골목에 흩어져 서서 식당을 향해 시선을 고정한 한국인 마을 청년들의 위협적인 전경.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S73sh1": {
  "input_fingerprint": "76229a380a658441",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.435,
     "y": 0.222
    },
    "look_target_anchor_en": "At the left flank of the giant ritual rock within the rock-and-altar forecourt.",
    "origin": {
     "x": 0.315,
     "y": 0.252
    },
    "origin_anchor_en": "On the outer left edge of the rock-and-altar forecourt beside the open clearing ground.",
    "view_left": {
     "x": 0.515,
     "y": 0.13
    },
    "view_left_anchor_en": "At the upper edge of the giant ritual rock within the forecourt.",
    "view_right": {
     "x": 0.535,
     "y": 0.305
    },
    "view_right_anchor_en": "On the clearing-side edge of the central altar forecourt."
   },
   "entity_placements": [
    {
     "anchor_en": "At the left flank of the giant ritual rock within the rock-and-altar forecourt.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.455,
      "y": 0.345
     },
     "facing_anchor_en": "On the open forecourt ground just beyond the clearing side of the central altar.",
     "slot": "E1",
     "subject_en": "white-haired woman raising ritual bells",
     "x": 0.435,
     "y": 0.222
    }
   ],
   "rationale_ko": "흰 머리 여자를 거대한 바위의 왼쪽 측면 순환 동선에 두고, 한 발을 내디디며 몸을 트는 방향을 제단 앞의 열린 공간 쪽으로 설정했다. 카메라는 동선 바깥쪽에서 측면으로 접근해 상체를 중간 크기로 포착하며, 좁은 화각 안에 인물과 거대한 바위 일부가 함께 들어오도록 배치했다. 모닥불은 지도에 고정 요소로 표시되지 않아 좌표 앵커로 사용하지 않았다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle On the outer left edge of the rock-and-altar forecourt beside the open clearing ground. (coarse estimate x=0.32, y=0.25); its sight line ends at LOOK TARGET: At the left flank of the giant ritual rock within the rock-and-altar forecourt. (coarse estimate x=0.43, y=0.22); the frame's LEFT edge cuts through At the upper edge of the giant ritual rock within the forecourt. (coarse estimate x=0.52, y=0.13) and its RIGHT edge through On the clearing-side edge of the central altar forecourt. (coarse estimate x=0.54, y=0.30)\n- E1 (white-haired woman raising ritual bells): circle At the left flank of the giant ritual rock within the rock-and-altar forecourt. (coarse estimate x=0.43, y=0.22), facing On the open forecourt ground just beyond the clearing side of the central altar. (coarse estimate x=0.46, y=0.34)\n\n[REGENERATE — the previous attempt failed the marker check: 카메라(CAM) 마커가 지정된 '바위와 제단 앞마당(Rock-and-Altar Forecourt)'의 왼쪽 가장자리가 아닌 완전히 다른 구역인 왼쪽 숲 가장자리(녹색 영역)에 잘못 배치되어 있습니다.. Draw exactly the specified markers and change nothing else on the plan.]",
  "marker_attempts": 2,
  "marker_map_sha256": "002cd16e4509ddb3c47a8328912f7ca8c4d699a331b44a2436b8b577bce220b3",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "기호, 텍스트 또는 평면도 요소가 발견되지 않았습니다. 입체적인 장면이 잘 그려져 있습니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands On the outer left edge of the rock-and-altar forecourt beside the open clearing ground. (coarse estimate x=0.32, y=0.25); its sight line ends at LOOK TARGET: At the left flank of the giant ritual rock within the rock-and-altar forecourt. (coarse estimate x=0.43, y=0.22); the frame's LEFT edge cuts through At the upper edge of the giant ritual rock within the forecourt. (coarse estimate x=0.52, y=0.13) and its RIGHT edge through On the clearing-side edge of the central altar forecourt. (coarse estimate x=0.54, y=0.30) — everything between those two edges is in frame, anything outside them is out of frame\n- white-haired woman raising ritual bells stands At the left flank of the giant ritual rock within the rock-and-altar forecourt. (coarse estimate x=0.43, y=0.22), facing On the open forecourt ground just beyond the clearing side of the central altar. (coarse estimate x=0.46, y=0.34) — from this camera the subject is near the CENTER of the frame and faces toward the RIGHT of the frame\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Track laterally at torso height along the outer edge of Youngran's path beside the large rock, tightening to a medium upper-body frame without changing the established movement direction. Catch her weight transferring into the turn, with the raised ritual bells held in the upper third, the bonfire offset behind her, and part of the rock anchoring the opposite edge.\n\nTHE LOCATION: 숲속 만월제 도당\n\nSHOT TEXT (authoritative, Korean): 타오르는 커다란 모닥불 주위 거대한 바위 옆에서 한 발을 내딛고 몸을 튼 채, 한 손으로 무령을 허공에 세차게 치켜든 강영란(흰 머리 여자)의 상체 자세.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S76sh3": {
  "input_fingerprint": "79f1649ed5590c16",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.57,
     "y": 0.53
    },
    "look_target_anchor_en": "On the woodland path southwest of the slaughterhouse entrance.",
    "origin": {
     "x": 0.47,
     "y": 0.61
    },
    "origin_anchor_en": "At the southern edge of the woodland path southwest of the slaughterhouse entrance.",
    "view_left": {
     "x": 0.54,
     "y": 0.465
    },
    "view_left_anchor_en": "At the northern edge of the woodland path west of the slaughterhouse entrance.",
    "view_right": {
     "x": 0.75,
     "y": 0.58
    },
    "view_right_anchor_en": "Near the southern edge of the woodland path east of the slaughterhouse entrance."
   },
   "entity_placements": [
    {
     "anchor_en": "On the woodland path diagonally southwest of the slaughterhouse entrance.",
     "camera_facing_evidence": "역동적인 뒷모습",
     "camera_facing_relation": "away_from_camera",
     "facing": {
      "x": 0.66,
      "y": 0.455
     },
     "facing_anchor_en": "At the path-side entrance into the slaughterhouse.",
     "slot": "E1",
     "subject_en": "pivoting running woman",
     "x": 0.56,
     "y": 0.55
    }
   ],
   "rationale_ko": "인물은 길에서 도축장 입구를 향해 대각선으로 방향을 틀며 한 발을 뻗는 순간에 배치했다. 카메라는 허리 높이의 후방 사선 위치에서 인물의 역동적인 뒷모습과 목적지인 입구를 함께 바라보며, 넓은 프레임 안에 길도 규모 기준으로 남긴다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle At the southern edge of the woodland path southwest of the slaughterhouse entrance. (coarse estimate x=0.47, y=0.61); its sight line ends at LOOK TARGET: On the woodland path southwest of the slaughterhouse entrance. (coarse estimate x=0.57, y=0.53); the frame's LEFT edge cuts through At the northern edge of the woodland path west of the slaughterhouse entrance. (coarse estimate x=0.54, y=0.47) and its RIGHT edge through Near the southern edge of the woodland path east of the slaughterhouse entrance. (coarse estimate x=0.75, y=0.58)\n- E1 (pivoting running woman): circle On the woodland path diagonally southwest of the slaughterhouse entrance. (coarse estimate x=0.56, y=0.55), facing At the path-side entrance into the slaughterhouse. (coarse estimate x=0.66, y=0.46)\n\n[REGENERATE — the previous attempt failed the marker check: 카메라의 시야각(웨지)이 잘못된 방향(서쪽)으로 열려 있으며, 피사체 E1이 시야각 바깥에 있습니다.. Draw exactly the specified markers and change nothing else on the plan.]",
  "marker_attempts": 2,
  "marker_map_sha256": "3b96bc37d27b46b11445a6ff730be676c576e59f53814f40375224c1a0de6022",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "도면 기호나 문자, 평면도 같은 데이터 유출이 없습니다. 스케치는 카메라 방향에 맞게 정상적으로 입체감 있는 장면을 보여줍니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands At the southern edge of the woodland path southwest of the slaughterhouse entrance. (coarse estimate x=0.47, y=0.61); its sight line ends at LOOK TARGET: On the woodland path southwest of the slaughterhouse entrance. (coarse estimate x=0.57, y=0.53); the frame's LEFT edge cuts through At the northern edge of the woodland path west of the slaughterhouse entrance. (coarse estimate x=0.54, y=0.47) and its RIGHT edge through Near the southern edge of the woodland path east of the slaughterhouse entrance. (coarse estimate x=0.75, y=0.58) — everything between those two edges is in frame, anything outside them is out of frame\n- pivoting running woman stands On the woodland path diagonally southwest of the slaughterhouse entrance. (coarse estimate x=0.56, y=0.55), facing At the path-side entrance into the slaughterhouse. (coarse estimate x=0.66, y=0.46) — from this camera the subject is near the CENTER of the frame and faces away from the camera, into the depth of the shot, so the camera sees this subject from behind\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Track at waist height behind and slightly beside Minsook in a wide, medium-full-body view as she pivots off the road and launches toward the slaughterhouse. Freeze the key instant with one foot airborne, her back driving diagonally from the lower foreground toward the entrance in the background, while the road remains visible as a scale reference.\n\nTHE LOCATION: 도축장 옆 숲길\n\nSHOT TEXT (authoritative, Korean): 어둠 속 달빛이 비치는 염소 도축장을 향해 방향을 틀고 급히 한 발을 허공에 뻗은 강민숙(홍은미)의 역동적인 뒷모습.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S78sh4": {
  "input_fingerprint": "ca361252e0be7b82",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.6,
     "y": 0.475
    },
    "look_target_anchor_en": "On the right half of the open clearing, midway between the altar forecourt and the forest edge.",
    "origin": {
     "x": 0.45,
     "y": 0.39
    },
    "origin_anchor_en": "On the upper open clearing just below and left of the altar forecourt.",
    "view_left": {
     "x": 0.69,
     "y": 0.47
    },
    "view_left_anchor_en": "On the right side of the open clearing below the altar forecourt.",
    "view_right": {
     "x": 0.62,
     "y": 0.56
    },
    "view_right_anchor_en": "On the middle-right open clearing toward the surrounding forest edge."
   },
   "entity_placements": [
    {
     "anchor_en": "On the upper open clearing just below the altar forecourt.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.65,
      "y": 0.515
     },
     "facing_anchor_en": "On the right half of the open clearing, midway between the altar forecourt and the forest edge.",
     "slot": "E1",
     "subject_en": "white-haired woman extending one hand",
     "x": 0.48,
     "y": 0.405
    },
    {
     "anchor_en": "On the right half of the open clearing, midway between the altar forecourt and the forest edge.",
     "camera_facing_evidence": "등을 돌린 고산들을 향해",
     "camera_facing_relation": "away_from_camera",
     "facing": {
      "x": 0.72,
      "y": 0.56
     },
     "facing_anchor_en": "On the outer-right open clearing beside the surrounding forest edge.",
     "slot": "E2",
     "subject_en": "departing man with his back turned",
     "x": 0.61,
     "y": 0.49
    }
   ],
   "rationale_ko": "흰 머리 여자는 제단 앞쪽의 상부 공터에서 오른쪽으로 떠나는 남자를 향해 손을 뻗는다. 카메라는 여자 곁에서 팔의 방향을 따라 바라보며, 여자는 근경 좌측에, 등을 돌린 남자는 더 작은 우측 배경에 놓이도록 구성했다. 남자의 진행 방향은 카메라 시선과 같은 깊이 방향이어서 뒷모습 조건과 일치한다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle On the upper open clearing just below and left of the altar forecourt. (coarse estimate x=0.45, y=0.39); its sight line ends at LOOK TARGET: On the right half of the open clearing, midway between the altar forecourt and the forest edge. (coarse estimate x=0.60, y=0.47); the frame's LEFT edge cuts through On the right side of the open clearing below the altar forecourt. (coarse estimate x=0.69, y=0.47) and its RIGHT edge through On the middle-right open clearing toward the surrounding forest edge. (coarse estimate x=0.62, y=0.56)\n- E1 (white-haired woman extending one hand): circle On the upper open clearing just below the altar forecourt. (coarse estimate x=0.48, y=0.41), facing On the right half of the open clearing, midway between the altar forecourt and the forest edge. (coarse estimate x=0.65, y=0.52)\n- E2 (departing man with his back turned): circle On the right half of the open clearing, midway between the altar forecourt and the forest edge. (coarse estimate x=0.61, y=0.49), facing On the outer-right open clearing beside the surrounding forest edge. (coarse estimate x=0.72, y=0.56)\n\n[REGENERATE — the previous attempt failed the marker check: 지정된 마커 외에 캐릭터를 설명하는 불필요한 텍스트가 추가되어 있습니다.. Draw exactly the specified markers and change nothing else on the plan.]",
  "marker_attempts": 3,
  "marker_map_sha256": "1cf00876eadc1b7564713383651fd1246b5fe07fb2d5328a522b0877f13bea90",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "도면 기호나 텍스트 누출이 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 MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands On the upper open clearing just below and left of the altar forecourt. (coarse estimate x=0.45, y=0.39); its sight line ends at LOOK TARGET: On the right half of the open clearing, midway between the altar forecourt and the forest edge. (coarse estimate x=0.60, y=0.47); the frame's LEFT edge cuts through On the right side of the open clearing below the altar forecourt. (coarse estimate x=0.69, y=0.47) and its RIGHT edge through On the middle-right open clearing toward the surrounding forest edge. (coarse estimate x=0.62, y=0.56) — everything between those two edges is in frame, anything outside them is out of frame\n- white-haired woman extending one hand stands On the upper open clearing just below the altar forecourt. (coarse estimate x=0.48, y=0.41), facing On the right half of the open clearing, midway between the altar forecourt and the forest edge. (coarse estimate x=0.65, y=0.52) — from this camera the subject is near the CENTER of the frame and faces away from the camera, into the depth of the shot, so the camera sees this subject from behind; this subject is turned toward the departing man with his back turned\n- departing man with his back turned stands On the right half of the open clearing, midway between the altar forecourt and the forest edge. (coarse estimate x=0.61, y=0.49), facing On the outer-right open clearing beside the surrounding forest edge. (coarse estimate x=0.72, y=0.56) — from this camera the subject is near the CENTER of the frame and faces away from the camera, into the depth of the shot, so the camera sees this subject from behind\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): At chest height beside Youngran, the camera finishes its inward move about one meter from her extended arm and looks along it toward Sandeul's turned back. Her hand and forearm occupy the near middle-left without exceeding the human-focused frame, while Sandeul remains smaller in the right background, preserving the physical line of interruption.\n\nTHE LOCATION: 숲속 만월제 도당\n\nSHOT TEXT (authoritative, Korean): 등을 돌린 고산들을 향해 단호하게 한 손을 뻗은 강영란(흰 머리 여자)의 팔.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S79sh1": {
  "input_fingerprint": "3c1758eb2faa605f",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.56,
     "y": 0.55
    },
    "look_target_anchor_en": "On the main island pier deck beside the water-contacting vessel berth edge.",
    "origin": {
     "x": 0.5,
     "y": 0.995
    },
    "origin_anchor_en": "In the sea directly beyond the seaward terminal end of the island pier.",
    "view_left": {
     "x": 0.18,
     "y": 0.47
    },
    "view_left_anchor_en": "At the left side of the pier-side building frontage beside the sea.",
    "view_right": {
     "x": 0.9,
     "y": 0.47
    },
    "view_right_anchor_en": "At the right side of the pier-side building frontage beside the sea."
   },
   "entity_placements": [
    {
     "anchor_en": "In the sea directly alongside the water-contacting vessel berth edge of the main island pier deck.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.66,
      "y": 0.52
     },
     "facing_anchor_en": "In the sea beside the village-side end of the water-contacting vessel berth edge.",
     "slot": "E1",
     "subject_en": "moored old fishing boat",
     "x": 0.66,
     "y": 0.69
    }
   ],
   "rationale_ko": "카메라는 외해 쪽 부두 끝 너머의 높은 위치에서 마을 방향으로 부두 전체를 길게 바라보도록 배치했다. 낡은 어선은 우측 접안 가장자리에 평행하게 정박시키고 화면 중앙보다 아래에 두어, 밝게 조명된 부두가 어선을 넓게 둘러싸는 원거리 전경이 되게 했다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle In the sea directly beyond the seaward terminal end of the island pier. (coarse estimate x=0.50, y=0.99); its sight line ends at LOOK TARGET: On the main island pier deck beside the water-contacting vessel berth edge. (coarse estimate x=0.56, y=0.55); the frame's LEFT edge cuts through At the left side of the pier-side building frontage beside the sea. (coarse estimate x=0.18, y=0.47) and its RIGHT edge through At the right side of the pier-side building frontage beside the sea. (coarse estimate x=0.90, y=0.47)\n- E1 (moored old fishing boat): circle In the sea directly alongside the water-contacting vessel berth edge of the main island pier deck. (coarse estimate x=0.66, y=0.69), facing In the sea beside the village-side end of the water-contacting vessel berth edge. (coarse estimate x=0.66, y=0.52)",
  "marker_attempts": 1,
  "marker_map_sha256": "5d6560ac4903d213a0a2d0b3c8eef1b0f2a3d79611682c9d835a6b3397379490",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "마커나 기호 없이 정상적인 투시도로 그려졌습니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands In the sea directly beyond the seaward terminal end of the island pier. (coarse estimate x=0.50, y=0.99); its sight line ends at LOOK TARGET: On the main island pier deck beside the water-contacting vessel berth edge. (coarse estimate x=0.56, y=0.55); the frame's LEFT edge cuts through At the left side of the pier-side building frontage beside the sea. (coarse estimate x=0.18, y=0.47) and its RIGHT edge through At the right side of the pier-side building frontage beside the sea. (coarse estimate x=0.90, y=0.47) — everything between those two edges is in frame, anything outside them is out of frame\n- moored old fishing boat stands In the sea directly alongside the water-contacting vessel berth edge of the main island pier deck. (coarse estimate x=0.66, y=0.69), facing In the sea beside the village-side end of the water-contacting vessel berth edge. (coarse estimate x=0.66, y=0.52) — from this camera the subject is in the RIGHT part of the frame and faces away from the camera, into the depth of the shot, so the camera sees this subject from behind\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): From an elevated position beyond the dock's outer edge, a static long-distance frame looks inward along the dock toward the moored fishing boat. The boat sits below center as a scale reference, surrounded by the brightly illuminated dock rather than isolated as an oversized foreground object.\n\nTHE LOCATION: 금월도 선착장\n\nSHOT TEXT (authoritative, Korean): 환한 조명등이 켜져 대낮처럼 밝은 선착장에 정박해 있는 낡은 낚시 어선의 외부 전경.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S85sh2": {
  "input_fingerprint": "3877c5843b4d0d0d",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.46,
     "y": 0.4
    },
    "look_target_anchor_en": "On the open village ground above the paved road and west of the fishing-gear warehouse.",
    "origin": {
     "x": 0.36,
     "y": 0.32
    },
    "origin_anchor_en": "On the open village ground southeast of the public-address loudspeaker installation.",
    "view_left": {
     "x": 0.43,
     "y": 0.36
    },
    "view_left_anchor_en": "On the open loudspeaker area east of the public-address installation.",
    "view_right": {
     "x": 0.46,
     "y": 0.45
    },
    "view_right_anchor_en": "On the village ground immediately above the paved road west of the fishing-gear warehouse."
   },
   "entity_placements": [
    {
     "anchor_en": "On the open village ground above the paved road and west of the fishing-gear warehouse.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.43,
      "y": 0.2
     },
     "facing_anchor_en": "Inside the uphill forest edge above the village boundary.",
     "slot": "E1",
     "subject_en": "watchful adult man",
     "x": 0.43,
     "y": 0.4
    }
   ],
   "rationale_ko": "인물은 숲으로 다시 올라가던 동선의 중간 지점에 두고 몸은 북쪽 숲을 향하게 했다. 카메라는 인물보다 숲 쪽으로 조금 앞서면서 옆에 위치해 상반신 미디엄 구도를 만들며, 인물을 화면 중간 왼쪽에 두도록 시선 중심을 약간 오른쪽으로 옮겨 창고 쪽을 의심스럽게 바라보는 여백을 확보했다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle On the open village ground southeast of the public-address loudspeaker installation. (coarse estimate x=0.36, y=0.32); its sight line ends at LOOK TARGET: On the open village ground above the paved road and west of the fishing-gear warehouse. (coarse estimate x=0.46, y=0.40); the frame's LEFT edge cuts through On the open loudspeaker area east of the public-address installation. (coarse estimate x=0.43, y=0.36) and its RIGHT edge through On the village ground immediately above the paved road west of the fishing-gear warehouse. (coarse estimate x=0.46, y=0.45)\n- E1 (watchful adult man): circle On the open village ground above the paved road and west of the fishing-gear warehouse. (coarse estimate x=0.43, y=0.40), facing Inside the uphill forest edge above the village boundary. (coarse estimate x=0.43, y=0.20)\n\n[REGENERATE — the previous attempt failed the marker check: 카메라 시야각(웨지)이 지정된 방향(동쪽)이 아닌 잘못된 방향(북쪽)을 향하고 있습니다.. Draw exactly the specified markers and change nothing else on the plan.]",
  "marker_attempts": 2,
  "marker_map_sha256": "ebda4931aa4c52e5da0c49bad8db19fada6b21ba1d675aaa648348272c1b0706",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "누출된 마커나 평면도 기호 없이 정상적인 3D 장면으로 그려졌습니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands On the open village ground southeast of the public-address loudspeaker installation. (coarse estimate x=0.36, y=0.32); its sight line ends at LOOK TARGET: On the open village ground above the paved road and west of the fishing-gear warehouse. (coarse estimate x=0.46, y=0.40); the frame's LEFT edge cuts through On the open loudspeaker area east of the public-address installation. (coarse estimate x=0.43, y=0.36) and its RIGHT edge through On the village ground immediately above the paved road west of the fishing-gear warehouse. (coarse estimate x=0.46, y=0.45) — everything between those two edges is in frame, anything outside them is out of frame\n- watchful adult man stands On the open village ground above the paved road and west of the fishing-gear warehouse. (coarse estimate x=0.43, y=0.40), facing Inside the uphill forest edge above the village boundary. (coarse estimate x=0.43, y=0.20) — from this camera the subject is near the CENTER of the frame and faces toward the LEFT of the frame and toward the camera, so the camera sees this subject from the front\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Track at chest height beside and slightly ahead of 고산들, holding a medium upper-body composition that preserves the contradiction between his forest-facing torso and his head turned toward the old gear warehouse. Place him on the middle-left, leaving open space on the right along his fixed warehouse-directed gaze so the suspicion reads geometrically before the camera moves closer.\n\nTHE LOCATION: 금월도 마을 스피커 탑 주변\n\nSHOT TEXT (authoritative, Korean): 몸의 방향은 숲을 향한 채 고개만 선착장 옆 낡은 어구 창고로 돌려 시선을 고정한 고산들의 상체.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S86sh5": {
  "input_fingerprint": "4a01b42bed94e300",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.515,
     "y": 0.345
    },
    "look_target_anchor_en": "On the village-side approach between the fishing-gear warehouse and Banwol Restaurant.",
    "origin": {
     "x": 0.5,
     "y": 0.68
    },
    "origin_anchor_en": "On the lower half of the main island pier deck.",
    "view_left": {
     "x": 0.22,
     "y": 0.14
    },
    "view_left_anchor_en": "In the woodland beside the western edge of the upper village houses.",
    "view_right": {
     "x": 0.8,
     "y": 0.14
    },
    "view_right_anchor_en": "In the woodland beside the eastern edge of the upper village houses."
   },
   "entity_placements": [
    {
     "anchor_en": "On the left half of the village-side approach just above the main pier deck.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.493,
      "y": 0.48
     },
     "facing_anchor_en": "At the village-side threshold of the main island pier deck.",
     "slot": "E1",
     "subject_en": "bent-forward sprinting Korean villager",
     "x": 0.493,
     "y": 0.405
    },
    {
     "anchor_en": "On the right half of the village-side approach just above the main pier deck.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.535,
      "y": 0.48
     },
     "facing_anchor_en": "At the village-side threshold of the main island pier deck.",
     "slot": "E2",
     "subject_en": "long-striding Korean villager",
     "x": 0.537,
     "y": 0.395
    },
    {
     "anchor_en": "In the narrow lane immediately left of the village-side approach and above the fishing-gear warehouse.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.485,
      "y": 0.41
     },
     "facing_anchor_en": "On the lower section of the village-side approach beside the fishing-gear warehouse frontage.",
     "slot": "E3",
     "subject_en": "charging Korean villager",
     "x": 0.455,
     "y": 0.305
    },
    {
     "anchor_en": "In the narrow lane immediately right of the village-side approach and above Banwol Restaurant.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.535,
      "y": 0.41
     },
     "facing_anchor_en": "On the lower section of the village-side approach beside the Banwol Restaurant frontage.",
     "slot": "E4",
     "subject_en": "bent-torso running Korean villager",
     "x": 0.585,
     "y": 0.3
    },
    {
     "anchor_en": "In the western narrow lane between the small village houses.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.485,
      "y": 0.34
     },
     "facing_anchor_en": "On the village-side approach just inland of the pier-side building frontage.",
     "slot": "E5",
     "subject_en": "descending Korean villager",
     "x": 0.376,
     "y": 0.225
    },
    {
     "anchor_en": "In the eastern narrow lane between the small village houses.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.54,
      "y": 0.34
     },
     "facing_anchor_en": "On the village-side approach just inland of the pier-side building frontage.",
     "slot": "E6",
     "subject_en": "sprinting Korean villager",
     "x": 0.654,
     "y": 0.205
    }
   ],
   "rationale_ko": "카메라는 주 부두 하단에서 마을 쪽을 넓게 올려다보도록 두었다. 선두 두 명은 부두 직전 접근로를 가로지르고, 후속 인원은 좌우 골목과 마을 주택 사이에서 연속적으로 내려오게 배치해 하나의 추격 흐름으로 보이게 했다. 전원은 화면 아래쪽의 보이지 않는 순찰정 방향으로 달린다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle On the lower half of the main island pier deck. (coarse estimate x=0.50, y=0.68); its sight line ends at LOOK TARGET: On the village-side approach between the fishing-gear warehouse and Banwol Restaurant. (coarse estimate x=0.52, y=0.34); the frame's LEFT edge cuts through In the woodland beside the western edge of the upper village houses. (coarse estimate x=0.22, y=0.14) and its RIGHT edge through In the woodland beside the eastern edge of the upper village houses. (coarse estimate x=0.80, y=0.14)\n- E1 (bent-forward sprinting Korean villager): circle On the left half of the village-side approach just above the main pier deck. (coarse estimate x=0.49, y=0.41), facing At the village-side threshold of the main island pier deck. (coarse estimate x=0.49, y=0.48)\n- E2 (long-striding Korean villager): circle On the right half of the village-side approach just above the main pier deck. (coarse estimate x=0.54, y=0.40), facing At the village-side threshold of the main island pier deck. (coarse estimate x=0.54, y=0.48)\n- E3 (charging Korean villager): circle In the narrow lane immediately left of the village-side approach and above the fishing-gear warehouse. (coarse estimate x=0.46, y=0.30), facing On the lower section of the village-side approach beside the fishing-gear warehouse frontage. (coarse estimate x=0.48, y=0.41)\n- E4 (bent-torso running Korean villager): circle In the narrow lane immediately right of the village-side approach and above Banwol Restaurant. (coarse estimate x=0.58, y=0.30), facing On the lower section of the village-side approach beside the Banwol Restaurant frontage. (coarse estimate x=0.54, y=0.41)\n- E5 (descending Korean villager): circle In the western narrow lane between the small village houses. (coarse estimate x=0.38, y=0.23), facing On the village-side approach just inland of the pier-side building frontage. (coarse estimate x=0.48, y=0.34)\n- E6 (sprinting Korean villager): circle In the eastern narrow lane between the small village houses. (coarse estimate x=0.65, y=0.20), facing On the village-side approach just inland of the pier-side building frontage. (coarse estimate x=0.54, y=0.34)",
  "marker_attempts": 1,
  "marker_map_sha256": "8d436b69581cf2ab099a8a6790fa7b29a25e51bf9fb116f099642f2e1b919714",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "도면 기호나 문자, 평면도 유출이 발견되지 않았습니다. 입체적인 장면 묘사입니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands On the lower half of the main island pier deck. (coarse estimate x=0.50, y=0.68); its sight line ends at LOOK TARGET: On the village-side approach between the fishing-gear warehouse and Banwol Restaurant. (coarse estimate x=0.52, y=0.34); the frame's LEFT edge cuts through In the woodland beside the western edge of the upper village houses. (coarse estimate x=0.22, y=0.14) and its RIGHT edge through In the woodland beside the eastern edge of the upper village houses. (coarse estimate x=0.80, y=0.14) — everything between those two edges is in frame, anything outside them is out of frame\n- bent-forward sprinting Korean villager stands On the left half of the village-side approach just above the main pier deck. (coarse estimate x=0.49, y=0.41), facing At the village-side threshold of the main island pier deck. (coarse estimate x=0.49, y=0.48) — from this camera the subject is near the CENTER of the frame and faces toward the camera, so the camera sees this subject from the front\n- long-striding Korean villager stands On the right half of the village-side approach just above the main pier deck. (coarse estimate x=0.54, y=0.40), facing At the village-side threshold of the main island pier deck. (coarse estimate x=0.54, y=0.48) — from this camera the subject is near the CENTER of the frame and faces toward the camera, so the camera sees this subject from the front\n- charging Korean villager stands In the narrow lane immediately left of the village-side approach and above the fishing-gear warehouse. (coarse estimate x=0.46, y=0.30), facing On the lower section of the village-side approach beside the fishing-gear warehouse frontage. (coarse estimate x=0.48, y=0.41) — from this camera the subject is near the CENTER of the frame and faces toward the RIGHT of the frame and toward the camera, so the camera sees this subject from the front; this subject is turned toward the bent-forward sprinting Korean villager\n- bent-torso running Korean villager stands In the narrow lane immediately right of the village-side approach and above Banwol Restaurant. (coarse estimate x=0.58, y=0.30), facing On the lower section of the village-side approach beside the Banwol Restaurant frontage. (coarse estimate x=0.54, y=0.41) — from this camera the subject is near the CENTER of the frame and faces toward the LEFT of the frame and toward the camera, so the camera sees this subject from the front; this subject is turned toward the bent-forward sprinting Korean villager\n- descending Korean villager stands In the western narrow lane between the small village houses. (coarse estimate x=0.38, y=0.23), facing On the village-side approach just inland of the pier-side building frontage. (coarse estimate x=0.48, y=0.34) — from this camera the subject is in the LEFT part of the frame and faces toward the RIGHT of the frame and toward the camera, so the camera sees this subject from the front; this subject is turned toward the charging Korean villager\n- sprinting Korean villager stands In the eastern narrow lane between the small village houses. (coarse estimate x=0.65, y=0.20), facing On the village-side approach just inland of the pier-side building frontage. (coarse estimate x=0.54, y=0.34) — from this camera the subject is in the RIGHT part of the frame and faces toward the LEFT of the frame and toward the camera, so the camera sees this subject from the front; this subject is turned toward the long-striding Korean villager\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): From slightly above head height near the lower dock, crane over a wide view looking back toward the upper side as the mass of Korean villagers pours downward. Arrange the leading bodies across the middle band with bent torsos and extended strides, while successive ranks climb into the upper background so the charge reads as a single predatory flow toward the unseen patrol boat below frame.\n\nTHE LOCATION: 금월도 선착장\n\nSHOT TEXT (authoritative, Korean): 선착장 위쪽에서 수리영 일행을 향해 몸을 굽히고 한 발을 뻗은 채 맹렬하게 뛰어내려오는 수많은 한국인 마을 사람들의 mid-action 전경.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S100sh5": {
  "input_fingerprint": "ff2e9d593b672cc2",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.68,
     "y": 0.4
    },
    "look_target_anchor_en": "On the main harbor dock just below its upper boat-facing berthing edge.",
    "origin": {
     "x": 0.24,
     "y": 0.52
    },
    "origin_anchor_en": "In the landside harbor access area just left of the main dock entrance.",
    "view_left": {
     "x": 0.5,
     "y": 0.2
    },
    "view_left_anchor_en": "In the harbor water beyond the upper boat-facing berthing edge.",
    "view_right": {
     "x": 0.77,
     "y": 0.54
    },
    "view_right_anchor_en": "On the outer half of the main harbor dock below its centerline."
   },
   "entity_placements": [
    {
     "anchor_en": "On the main harbor dock just below its upper boat-facing berthing edge.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.3,
      "y": 0.47
     },
     "facing_anchor_en": "At the inland entrance of the main harbor dock beside the landing sign.",
     "slot": "E1",
     "subject_en": "older fisherman with one arm raised overhead",
     "x": 0.68,
     "y": 0.4
    },
    {
     "anchor_en": "At the inland entrance of the main harbor dock beside the landing sign.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.68,
      "y": 0.4
     },
     "facing_anchor_en": "On the main harbor dock just below its upper boat-facing berthing edge.",
     "slot": "E2",
     "subject_en": "casually dressed man watching the fisherman",
     "x": 0.3,
     "y": 0.47
    }
   ],
   "rationale_ko": "카메라는 육상 접근부에서 선착장 길이 방향으로 바라보며, 입구의 남성 어깨가 프레임 왼쪽 전경에 살짝 걸리도록 배치했다. 손을 든 어부는 어선이 접안할 수 있는 위쪽 계류 가장자리 안쪽의 중우측에 전신으로 서고, 두 인물은 서로를 향해 충분한 거리를 두고 마주 본다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle In the landside harbor access area just left of the main dock entrance. (coarse estimate x=0.24, y=0.52); its sight line ends at LOOK TARGET: On the main harbor dock just below its upper boat-facing berthing edge. (coarse estimate x=0.68, y=0.40); the frame's LEFT edge cuts through In the harbor water beyond the upper boat-facing berthing edge. (coarse estimate x=0.50, y=0.20) and its RIGHT edge through On the outer half of the main harbor dock below its centerline. (coarse estimate x=0.77, y=0.54)\n- E1 (older fisherman with one arm raised overhead): circle On the main harbor dock just below its upper boat-facing berthing edge. (coarse estimate x=0.68, y=0.40), facing At the inland entrance of the main harbor dock beside the landing sign. (coarse estimate x=0.30, y=0.47)\n- E2 (casually dressed man watching the fisherman): circle At the inland entrance of the main harbor dock beside the landing sign. (coarse estimate x=0.30, y=0.47), facing On the main harbor dock just below its upper boat-facing berthing edge. (coarse estimate x=0.68, y=0.40)",
  "marker_attempts": 1,
  "marker_map_sha256": "1ec9c2870abb7b9eaae33673ddb6d4031fad9978ec15d2732c9a85811e6a77a1",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "도면 기호나 텍스트가 없는 정상적인 투시도입니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands In the landside harbor access area just left of the main dock entrance. (coarse estimate x=0.24, y=0.52); its sight line ends at LOOK TARGET: On the main harbor dock just below its upper boat-facing berthing edge. (coarse estimate x=0.68, y=0.40); the frame's LEFT edge cuts through In the harbor water beyond the upper boat-facing berthing edge. (coarse estimate x=0.50, y=0.20) and its RIGHT edge through On the outer half of the main harbor dock below its centerline. (coarse estimate x=0.77, y=0.54) — everything between those two edges is in frame, anything outside them is out of frame\n- older fisherman with one arm raised overhead stands On the main harbor dock just below its upper boat-facing berthing edge. (coarse estimate x=0.68, y=0.40), facing At the inland entrance of the main harbor dock beside the landing sign. (coarse estimate x=0.30, y=0.47) — from this camera the subject is near the CENTER of the frame and faces toward the camera, so the camera sees this subject from the front; this subject is turned toward the casually dressed man watching the fisherman\n- casually dressed man watching the fisherman stands At the inland entrance of the main harbor dock beside the landing sign. (coarse estimate x=0.30, y=0.47), facing On the main harbor dock just below its upper boat-facing berthing edge. (coarse estimate x=0.68, y=0.40) — from this camera the subject is in the LEFT part of the frame and faces away from the camera, into the depth of the shot, so the camera sees this subject from behind; this subject is turned toward the older fisherman with one arm raised overhead\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Let the flow's pan settle at eye level near 구창모's group, using a sliver of 구창모's shoulder at the left foreground to motivate the long view across the dock. 황봉업 appears full-length in the middle-right before the fishing boat, one arm raised overhead and his body turned toward 구창모, with surrounding dock space preserving the intended distance.\n\nTHE LOCATION: 육지 해안 마을 선착장\n\nSHOT TEXT (authoritative, Korean): 낚시 어선 앞에 서서 구창모를 향해 한 손을 머리 위로 번쩍 들어 올린 채 서 있는 황봉업의 전신.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S105sh1": {
  "input_fingerprint": "26a404415ccd5afb",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.5,
     "y": 0.54
    },
    "look_target_anchor_en": "At the center of the open clearing ground.",
    "origin": {
     "x": 0.5,
     "y": 0.76
    },
    "origin_anchor_en": "Over the southern outer edge of the open clearing ground near the surrounding forest edge.",
    "view_left": {
     "x": 0.19,
     "y": 0.44
    },
    "view_left_anchor_en": "At the western forest edge beside the open clearing ground.",
    "view_right": {
     "x": 0.81,
     "y": 0.44
    },
    "view_right_anchor_en": "At the eastern forest edge beside the open clearing ground."
   },
   "entity_placements": [
    {
     "anchor_en": "On the northern arc of the open clearing ground below the rock-and-altar forecourt.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.5,
      "y": 0.54
     },
     "facing_anchor_en": "At the center of the open clearing ground.",
     "slot": "E1",
     "subject_en": "black-clad devotee arc",
     "x": 0.5,
     "y": 0.42
    },
    {
     "anchor_en": "On the northeastern arc of the open clearing ground.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.5,
      "y": 0.54
     },
     "facing_anchor_en": "At the center of the open clearing ground.",
     "slot": "E2",
     "subject_en": "black-clad devotee arc",
     "x": 0.64,
     "y": 0.48
    },
    {
     "anchor_en": "On the southeastern arc of the open clearing ground.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.5,
      "y": 0.54
     },
     "facing_anchor_en": "At the center of the open clearing ground.",
     "slot": "E3",
     "subject_en": "black-clad devotee arc",
     "x": 0.64,
     "y": 0.6
    },
    {
     "anchor_en": "On the southern arc of the open clearing ground inside the surrounding forest edge.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.5,
      "y": 0.54
     },
     "facing_anchor_en": "At the center of the open clearing ground.",
     "slot": "E4",
     "subject_en": "black-clad devotee arc",
     "x": 0.5,
     "y": 0.66
    },
    {
     "anchor_en": "On the southwestern arc of the open clearing ground.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.5,
      "y": 0.54
     },
     "facing_anchor_en": "At the center of the open clearing ground.",
     "slot": "E5",
     "subject_en": "black-clad devotee arc",
     "x": 0.36,
     "y": 0.6
    },
    {
     "anchor_en": "On the northwestern arc of the open clearing ground.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.5,
      "y": 0.54
     },
     "facing_anchor_en": "At the center of the open clearing ground.",
     "slot": "E6",
     "subject_en": "black-clad devotee arc",
     "x": 0.36,
     "y": 0.48
    }
   ],
   "rationale_ko": "신도 수십 명을 여섯 개의 군중 호로 나누어 공터 중앙을 완전히 둘러싸는 원형 대형으로 배치했다. 모든 호는 중앙 모닥불 위치를 향하며, 카메라는 남쪽 대형 바깥의 높은 크레인 위치에서 넓은 하향 시야로 전체 원과 주변 의식 공간을 담는다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle Over the southern outer edge of the open clearing ground near the surrounding forest edge. (coarse estimate x=0.50, y=0.76); its sight line ends at LOOK TARGET: At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54); the frame's LEFT edge cuts through At the western forest edge beside the open clearing ground. (coarse estimate x=0.19, y=0.44) and its RIGHT edge through At the eastern forest edge beside the open clearing ground. (coarse estimate x=0.81, y=0.44)\n- E1 (black-clad devotee arc): circle On the northern arc of the open clearing ground below the rock-and-altar forecourt. (coarse estimate x=0.50, y=0.42), facing At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54)\n- E2 (black-clad devotee arc): circle On the northeastern arc of the open clearing ground. (coarse estimate x=0.64, y=0.48), facing At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54)\n- E3 (black-clad devotee arc): circle On the southeastern arc of the open clearing ground. (coarse estimate x=0.64, y=0.60), facing At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54)\n- E4 (black-clad devotee arc): circle On the southern arc of the open clearing ground inside the surrounding forest edge. (coarse estimate x=0.50, y=0.66), facing At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54)\n- E5 (black-clad devotee arc): circle On the southwestern arc of the open clearing ground. (coarse estimate x=0.36, y=0.60), facing At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54)\n- E6 (black-clad devotee arc): circle On the northwestern arc of the open clearing ground. (coarse estimate x=0.36, y=0.48), facing At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54)",
  "marker_attempts": 1,
  "marker_map_sha256": "866af1e90363c65885bf3130e3af1c13e84316b5d262fcd4fb7d86e6b2e50617",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "마커나 다이어그램 기호가 없는 정상적인 스케치입니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands Over the southern outer edge of the open clearing ground near the surrounding forest edge. (coarse estimate x=0.50, y=0.76); its sight line ends at LOOK TARGET: At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54); the frame's LEFT edge cuts through At the western forest edge beside the open clearing ground. (coarse estimate x=0.19, y=0.44) and its RIGHT edge through At the eastern forest edge beside the open clearing ground. (coarse estimate x=0.81, y=0.44) — everything between those two edges is in frame, anything outside them is out of frame\n- black-clad devotee arc stands On the northern arc of the open clearing ground below the rock-and-altar forecourt. (coarse estimate x=0.50, y=0.42), facing At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54) — from this camera the subject is near the CENTER of the frame and faces toward the camera, so the camera sees this subject from the front\n- black-clad devotee arc stands On the northeastern arc of the open clearing ground. (coarse estimate x=0.64, y=0.48), facing At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54) — from this camera the subject is in the RIGHT part of the frame and faces toward the LEFT of the frame and toward the camera, so the camera sees this subject from the front\n- black-clad devotee arc stands On the southeastern arc of the open clearing ground. (coarse estimate x=0.64, y=0.60), facing At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54) — from this camera the subject is in the RIGHT part of the frame and faces toward the LEFT of the frame and away from the camera, into the depth of the shot, so the camera sees this subject from behind\n- black-clad devotee arc stands On the southern arc of the open clearing ground inside the surrounding forest edge. (coarse estimate x=0.50, y=0.66), facing At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54) — from this camera the subject is near the CENTER of the frame and faces away from the camera, into the depth of the shot, so the camera sees this subject from behind\n- black-clad devotee arc stands On the southwestern arc of the open clearing ground. (coarse estimate x=0.36, y=0.60), facing At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54) — from this camera the subject is in the LEFT part of the frame and faces toward the RIGHT of the frame and away from the camera, into the depth of the shot, so the camera sees this subject from behind\n- black-clad devotee arc stands On the northwestern arc of the open clearing ground. (coarse estimate x=0.36, y=0.48), facing At the center of the open clearing ground. (coarse estimate x=0.50, y=0.54) — from this camera the subject is in the LEFT part of the frame and faces toward the RIGHT of the frame and toward the camera, so the camera sees this subject from the front\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): Crane high above the outer edge of the gathering and angle steeply downward, revealing the complete circular formation of Korean black-clad devotees around the large bonfire. The fire occupies less than the central third while the human ring and surrounding ritual ground spread across the wider frame, emphasizing collective scale and isolation.\n\nTHE LOCATION: 숲속 만월제 도당\n\nSHOT TEXT (authoritative, Korean): 숲속 도당, 활활 치솟는 커다란 모닥불 주변을 둥글게 둘러싼 수십 명의 한국인 검은 옷 신도들의 압도적인 전경.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S95sh15": {
  "input_fingerprint": "67dfd5d7444c392a",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.67,
     "y": 0.65
    },
    "look_target_anchor_en": "Over the high cliff drop just seaward of the brink on the middle-right section of the overlook.",
    "origin": {
     "x": 0.7,
     "y": 0.43
    },
    "origin_anchor_en": "On the cliff-top standing ground inland of the middle-right section of the brink.",
    "view_left": {
     "x": 0.82,
     "y": 0.73
    },
    "view_left_anchor_en": "Among the rocks and water at the foot of the cliff on the right side of the lower reef.",
    "view_right": {
     "x": 0.47,
     "y": 0.73
    },
    "view_right_anchor_en": "In the sea among the smaller rocks at the foot of the central cliff drop."
   },
   "entity_placements": [
    {
     "anchor_en": "Suspended over the high cliff drop just seaward of the middle-right brink.",
     "camera_facing_evidence": "",
     "camera_facing_relation": "unspecified",
     "facing": {
      "x": 0.62,
      "y": 0.52
     },
     "facing_anchor_en": "On the cliff-top standing ground immediately inland of the middle-right brink.",
     "slot": "E1",
     "subject_en": "airborne injured man",
     "x": 0.62,
     "y": 0.63
    }
   ],
   "rationale_ko": "인물은 발길질 직후 단단한 지면을 벗어나 절벽 턱 바깥의 낙하 공간에 떠 있는 순간으로 배치했다. 몸의 정면은 발길질을 받은 절벽 위쪽을 향하게 하여 뒤로 밀려나는 동작을 표현했다. 카메라는 절벽 안쪽 허리 높이 위치에서 낙하 방향을 바라보며, 인물을 화면 중우측에 두고 절벽 턱과 아래의 열린 공간 및 암초까지 폭넓게 포함한다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle On the cliff-top standing ground inland of the middle-right section of the brink. (coarse estimate x=0.70, y=0.43); its sight line ends at LOOK TARGET: Over the high cliff drop just seaward of the brink on the middle-right section of the overlook. (coarse estimate x=0.67, y=0.65); the frame's LEFT edge cuts through Among the rocks and water at the foot of the cliff on the right side of the lower reef. (coarse estimate x=0.82, y=0.73) and its RIGHT edge through In the sea among the smaller rocks at the foot of the central cliff drop. (coarse estimate x=0.47, y=0.73)\n- E1 (airborne injured man): circle Suspended over the high cliff drop just seaward of the middle-right brink. (coarse estimate x=0.62, y=0.63), facing On the cliff-top standing ground immediately inland of the middle-right brink. (coarse estimate x=0.62, y=0.52)",
  "marker_attempts": 1,
  "marker_map_sha256": "f31995bb13b94329dba1a08d15bff093e272be0890c021ef43bd1df889411a49",
  "leakage": {
   "has_marker_leakage": false,
   "details_ko": "스케치에 마커나 기호 등의 평면도 요소가 없습니다."
  },
  "prompt": "Draw a single CAMERA-VIEW storyboard frame (16:9) of the scene staged\non the attached MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands On the cliff-top standing ground inland of the middle-right section of the brink. (coarse estimate x=0.70, y=0.43); its sight line ends at LOOK TARGET: Over the high cliff drop just seaward of the brink on the middle-right section of the overlook. (coarse estimate x=0.67, y=0.65); the frame's LEFT edge cuts through Among the rocks and water at the foot of the cliff on the right side of the lower reef. (coarse estimate x=0.82, y=0.73) and its RIGHT edge through In the sea among the smaller rocks at the foot of the central cliff drop. (coarse estimate x=0.47, y=0.73) — everything between those two edges is in frame, anything outside them is out of frame\n- airborne injured man stands Suspended over the high cliff drop just seaward of the middle-right brink. (coarse estimate x=0.62, y=0.63), facing On the cliff-top standing ground immediately inland of the middle-right brink. (coarse estimate x=0.62, y=0.52) — from this camera the subject is in the RIGHT part of the frame and faces toward the camera, so the camera sees this subject from the front\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): At waist height and parallel to the cliff edge, the tracking camera begins wide enough to hold 고산들's entire body as the kick carries him backward beyond solid ground. His airborne body tilts across the middle-right of frame, with the cliff edge retained below and open space visibly extending behind him so the direction and danger of the fall are unmistakable.\n\nTHE LOCATION: 금월도 해안 절벽\n\nSHOT TEXT (authoritative, Korean): 강한 발길질의 충격으로 절벽 밖 허공에 붕 뜬 채 뒤로 기우는 고산들의 몸.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 },
 "S104sh4": {
  "input_fingerprint": "903b916f89c7d2f5",
  "geometry": {
   "camera": {
    "look_target": {
     "x": 0.5,
     "y": 0.815
    },
    "look_target_anchor_en": "On the village-road centerline below the restaurant-front alley.",
    "origin": {
     "x": 0.5,
     "y": 0.735
    },
    "origin_anchor_en": "On the restaurant-front alley centerline at the village-road approach.",
    "view_left": {
     "x": 0.46,
     "y": 0.86
    },
    "view_left_anchor_en": "On the left half of the uphill village road above the forest threshold.",
    "view_right": {
     "x": 0.54,
     "y": 0.86
    },
    "view_right_anchor_en": "On the right half of the uphill village road above the forest threshold."
   },
   "entity_placements": [
    {
     "anchor_en": "On the village-road centerline below the restaurant-front alley.",
     "camera_facing_evidence": "위압적인 전면",
     "camera_facing_relation": "toward_camera",
     "facing": {
      "x": 0.5,
      "y": 0.68
     },
     "facing_anchor_en": "On the restaurant-front alley centerline directly before the front entrance.",
     "slot": "E1",
     "subject_en": "soaked advancing man",
     "x": 0.5,
     "y": 0.815
    }
   ],
   "rationale_ko": "인물을 식당으로 이어지는 마을길 중심축에 두고, 식당 입구 쪽을 향해 전진하도록 배치했다. 카메라는 인물과 입구 사이에서 같은 축을 따라 후퇴하며 정면을 담고, 좁은 화각으로 전신을 위압적으로 포착한다."
  },
  "marker_check": {
   "markers_ok": true,
   "details_ko": "정상"
  },
  "marker_prompt": "Edit the attached TOP-DOWN SITE PLAN image: draw STAGING MARKERS on it\nat the positions described below. This is a planning aid for a film\ncrew, drawn over the plan.\n\nPLACEMENT AUTHORITY: each marker line below places its marker relative\nto a fixed feature, zone or boundary VISIBLE on the attached plan —\nthat landmark-relative description is the placement authority. The\nnormalized fractions in parentheses are coarse estimates of the same\nspot (x runs left→right 0..1, y runs top→bottom 0..1 of the full\ncanvas) — use them only to confirm the rough region; when in doubt,\nthe landmark description wins. Do NOT relocate a marker onto a\ndifferent salient line or feature than the one its description names.\n\nNEVER move a marker to make the drawing fit. The camera circle and\nevery subject circle go exactly where their descriptions place them,\neven when that leaves the view wedge or an arrow running off the edge\nof the image — in that case just let the wedge or arrow continue to\nthe image border and stop there. Shifting the camera outward, pulling\na subject aside, or shrinking the wedge to keep it inside the canvas\nall break the plan this map exists to record.\n\nDraw EXACTLY these markers, flat and high-contrast, small enough not\nto hide the plan under them:\n- CAMERA: one filled circle at the camera position, labelled CAM, and\n  out of that circle a FIELD-OF-VIEW WEDGE showing what this shot\n  sees — two straight thin lines, one running toward each of the two\n  frame-edge positions given below, and a third thinner line to the\n  look-target position between them. Fill the area between the two\n  edge lines with a light translucent tint so the wedge reads as the\n  camera's view; the plan underneath must stay fully readable through\n  the tint. Every subject listed below stands inside this wedge —\n  if your drawing leaves one outside it, the wedge is drawn wrong,\n  not the subject.\n- SUBJECTS: one filled circle per subject at its position, labelled\n  with its slot code (E1, E2, ...) followed by its short description\n  in parentheses, and one short ARROW leaving that circle toward the\n  spot the subject faces, so the direction each subject is turned is\n  readable at a glance.\n\nChange NOTHING else: keep every zone, landmark, line, label and color\nof the plan exactly as attached — do not redraw, recolor, move or\nremove anything, do not add scenery, shadows or any other symbols or\ntext beyond the markers, wedge, arrows and labels specified above. The\noutput is the SAME plan image with only these staging markers added.\n\nMARKERS TO DRAW:\n- CAM (camera): circle On the restaurant-front alley centerline at the village-road approach. (coarse estimate x=0.50, y=0.73); its sight line ends at LOOK TARGET: On the village-road centerline below the restaurant-front alley. (coarse estimate x=0.50, y=0.81); the frame's LEFT edge cuts through On the left half of the uphill village road above the forest threshold. (coarse estimate x=0.46, y=0.86) and its RIGHT edge through On the right half of the uphill village road above the forest threshold. (coarse estimate x=0.54, y=0.86)\n- E1 (soaked advancing man): circle On the village-road centerline below the restaurant-front alley. (coarse estimate x=0.50, y=0.81), facing On the restaurant-front alley centerline directly before the front entrance. (coarse estimate x=0.50, y=0.68)\n\n[REGENERATE — the previous attempt failed the marker check: 카메라 마커의 위치가 잘못되었으며, 시야각이 카메라가 아닌 피사체(E1)에서 시작되어 반대 방향으로 열려 있습니다.. Draw exactly the specified markers and change nothing else on the plan.]",
  "marker_attempts": 2,
  "marker_map_sha256": "790f2654511c0105d40f6ae75de45434861a1f5f7b00b8d7c5550843619eaeef",
  "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 MARKER-ANNOTATED TOP-DOWN SITE PLAN, seen from the\ncamera position marked on it.\n\nThe attached plan carries STAGING MARKERS drawn over it: a\ncamera-position marker with a field-of-view wedge (where the camera\nstands, which way it looks and how wide it sees) and one labelled\ncircle per subject with an arrow (where each subject stands and which\nway that subject is turned). Jurisdiction of sources — follow all three, they do not\nconflict:\n- The SITE PLAN under the markers is the SOT for what exists on the\n  site and how its zones and landmarks are arranged relative to each\n  other. The markers themselves are staging aids, NOT site content.\n- The markers on the plan are the SOT for XY POSITIONS, horizontal\n  heading and body facing ONLY: where each subject stands on that\n  plan, which way each one is turned, where the camera stands, which\n  way it looks and how wide its wedge opens. Distances, directions\n  and relative depths between subjects come from the marked\n  positions; what the wedge covers is what this frame contains, and\n  subjects outside the wedge are out of frame. The STAGING GEOMETRY\n  text below restates the same staging with coarse normalized\n  estimates — it never overrides the marked plan.\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. It does NOT govern horizontal\n  placement: which side of the frame a subject occupies and which way\n  that subject is turned come from the marked plan and the staging\n  text, even when that memo also mentions left, right, foreground or\n  background.\n\nSTYLE — MANNEQUIN BLOCKING SKETCH: thin uniform dark outlines on plain\npaper, nothing else. Draw every person as a PLAIN ARTIST'S MANNEQUIN:\na featureless oval head with NO face, NO hair, NO clothing and no\ncostume detail — simple tapered limbs and torso, like a wooden posing\nfigure. Because the head has no face and no hair, the ONLY thing that\nshows which way a figure is turned is the BODY ITSELF: the direction\nthe chest, hips and feet point, and the way the head-oval sits on the\nshoulders. Aim each mannequin's chest and feet at the spot the staging\ntext says that person faces, and let the arms carry the action. A\nmannequin turned to the wrong side is the single worst error in this\ndrawing. Environment is reduced to a few placement lines (ground line,\nkey landmark silhouettes) — just enough to read positions and depth.\n\nSTAGING GEOMETRY — PLANNING DATA ONLY, NEVER RENDER IT: the lines\nbelow restate, marker by marker, the staging already marked on the\nattached plan: each landmark-relative description restates the\nmarker-map authority in words, and the normalized fractions in\nparentheses (x left→right 0..1, y top→bottom 0..1) are coarse audit\nestimates only — neither ever overrides the marked plan.\n\nThose fractions are TOP-DOWN bookkeeping. Never read them as left and\nright of your frame: on a plan seen from above, a subject \"to the\nright\" can sit on either side of a camera that looks any direction.\nEach subject line therefore ends with the SAME staging expressed from\nTHIS camera — which part of the frame the subject occupies, and which\nway that subject faces in screen terms (toward the left or right of\nthe frame, toward or away from the camera or side-on to it in\nprofile, and whom they are looking at). Those screen-space statements\nare what your drawing must match:\na figure said to face the right of the frame is turned to the right in\nyour drawing, never mirrored.\n\nDo NOT draw any of this data in the image — no numbers, no\ncoordinates, no labels, no arrows, no cones, no dots, no diagram\nsymbols of any kind.\n\nWhere the CAMERA DIRECTION text below also describes horizontal\nplacement (which side of the frame someone occupies, foreground left,\nbackground right), treat that as background reading only: for anything\nabout left, right and where a subject sits in this frame, the staging\nabove wins. CAMERA DIRECTION governs the vertical angle, elevation,\ntilt and framing.\n\nFor whether a subject faces the camera or is turned away from it, each\nsubject line above already states it in screen terms and the SHOT TEXT\nis the source it was derived from — draw exactly what the subject line\nsays, and never mirror it.\n- the camera stands On the restaurant-front alley centerline at the village-road approach. (coarse estimate x=0.50, y=0.73); its sight line ends at LOOK TARGET: On the village-road centerline below the restaurant-front alley. (coarse estimate x=0.50, y=0.81); the frame's LEFT edge cuts through On the left half of the uphill village road above the forest threshold. (coarse estimate x=0.46, y=0.86) and its RIGHT edge through On the right half of the uphill village road above the forest threshold. (coarse estimate x=0.54, y=0.86) — everything between those two edges is in frame, anything outside them is out of frame\n- soaked advancing man stands On the village-road centerline below the restaurant-front alley. (coarse estimate x=0.50, y=0.81), facing On the restaurant-front alley centerline directly before the front entrance. (coarse estimate x=0.50, y=0.68) — from this camera the subject is near the CENTER of the frame and faces toward the camera, so the camera sees this subject from the front\n\nCAMERA DIRECTION (vertical angle / elevation / framing SOT): At chest height on the road-to-entrance axis, retreat directly ahead of 고산들 in a tightened full-body frame, placing his hostile face in the upper center and his lifted stepping foot just above the lower edge. His soaked body advances squarely into the lens path while the restaurant entrance remains behind the camera, matching his fixed line of attack toward 황봉업.\n\nTHE LOCATION: 반월 식당 앞과 골목\n\nSHOT TEXT (authoritative, Korean): 흠뻑 젖은 모습으로 살기 어린 눈빛을 띤 채, 마을 길을 걷는 중간 한 발이 공중에 떠 있는 고산들의 위압적인 전면.\n\nNATURAL PERFORMANCE (default only — every explicit direction above\nwins): when no pose contract or shot-text direction says otherwise,\nsketched figures read as alive in mid-moment — believable weight\nshift, hands engaged, head-oval and chest turned toward the target the\nSHOT TEXT implies; avoid a stiff attention stance (feet together, arms\nhanging straight down) and a head-oval squared flat to the lens. If the\nSHOT TEXT or any POSE contract above stages stillness, death, sleep,\nunconsciousness, restraint, drill or an explicit direct-to-camera look,\ndraw THAT exactly — this clause never overrides it.\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."
 }
}