{
  "status": "completed",
  "completed_count": 24,
  "applicable_count": 24,
  "failed_count": 0,
  "schema_version": 13,
  "config_hash": "7f9c6632cec70179",
  "data": {
    "scenes": [
      {
        "scene_index": 1,
        "heading": "S1. INT. 아케론 딥 볼트 12, 폐광 갱도 — 밤 — 2026",
        "beat_title": "호기심에 이끌린 접촉 시도",
        "representative_moment": "토니(C01)가 옅은 미소를 띠며 빛을 흡수하는 검은 광맥을 향해 손을 뻗는다.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "Medium framing from upper-chest height, deep low-key high contrast lighting with selective helmet light illumination",
            "t2i_prompt": "Photorealistic cinematic still. Seen from upper-chest height, C01O01 in a black rescue workwear jacket, a Caucasian man in his 30s, occupies the left half of the frame at three-quarter angle from the right. He is leaning forward with an extended reach toward the right. His outstretched hand is fully visible between his faintly smiling face and a black glass-like mineral vein running along the right frame edge. His eyes are looking off-screen to the right, following the vein's path rather than the lens. A single helmet light cuts through the deep, low-key dark tunnel with high contrast, illuminating his upper body and the rugged, mineral-black coal layers forming a dark spatial boundary behind his reaching hand. The black mineral vein appears to swallow the illumination reaching its surface.",
            "outfit_assignments": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "source_facts": [
              "Tony reaches his hand toward the black vein with a faint smile",
              "The black vein absorbs light, located between coal layers",
              "At Tony's upper-chest height, view from his front-side three-quarter angle",
              "Tony left of center and the black vein along the right edge"
            ],
            "visual_inferences": [
              "fixed_element abandoned_mine_tunnel_environment: rugged, mineral-black rock surfaces and muted industrial textures",
              "Tony's outstretched arm bridges the visual gap between his face and the vein on the right side of the frame"
            ],
            "creative_decisions": [
              "Framed tightly from upper-chest height to ensure both the facial expression and the reaching hand are clearly readable in the same composition",
              "Emphasized the high-contrast lighting to highlight the physical texture of the coal layers against the light-absorbing black vein"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [],
            "reference_phrase_kinds": [
              "character",
              "background"
            ],
            "owned_object_usage": [
              {
                "owned_token": "blast door",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "ceiling light",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "drainage channel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "mineral vein",
                "usage_kind": "anchor",
                "source_phrase": "a black glass-like mineral vein running along the right frame edge"
              },
              {
                "owned_token": "rail",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "sensor",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "support frame",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tunnel",
                "usage_kind": "anchor",
                "source_phrase": "the deep, low-key dark tunnel with high contrast"
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "279ca780b5ea189f",
              "owned_hash": "c96656d7ec4f6bb6",
              "camera_direction_hash": "5a5d9b3424a748ca",
              "owned_usage_hash": "36168f0ca72d8c4a",
              "validator": "scene_detail_owned_objects.v1",
              "violations": [
                {
                  "owned_object": "mineral vein",
                  "violating_phrase": "a black glass-like mineral vein running along the right frame edge",
                  "reason": "The mineral vein is used as a spatial composition anchor and lighting reference without requesting a new or redesigned object, matching the producer's anchor declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "tunnel",
                  "violating_phrase": "the deep, low-key dark tunnel with high contrast",
                  "reason": "The tunnel is mentioned solely as existing environmental and lighting context, matching the producer's anchor declaration.",
                  "verdict": "anchor_reference"
                }
              ]
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": -1,
        "dependency_reason": "N/A",
        "_shot_index": 8,
        "visible_entities": [
          "C01",
          "L01"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 1,
            "shot_index": 8
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "medium",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "At Tony's upper-chest height, the dolly-in settles into a tight upper-body view from his front-side three-quarter angle, placing 토니(앤서니 로저스) left of center and the black vein along the right edge of his reaching path. His extended hand remains fully visible between his faint smile and the vein, while his eyes follow the vein toward its off-screen continuation rather than the lens.",
            "lighting_mood": "The single helmet light cuts through the dark tunnel in restrained low-key contrast while the black vein appears to swallow the illumination reaching it.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 1명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=medium — close 규칙은 이 샷에 해당하지 않는다"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": null,
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "C01",
              "L01"
            ],
            "allowed_outlook_pairs": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "background_ref_attached",
            "bg_id": "L01B03",
            "reference_usage": "exact_background",
            "owned_objects": [
              "blast door",
              "ceiling light",
              "drainage channel",
              "mineral vein",
              "rail",
              "sensor",
              "support frame",
              "tunnel"
            ],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "owned_objects lists all environment objects already drawn in the background reference — do not create new ones",
              "if focusing on an owned object, declare its reference-kind via the reference_phrase_kinds sidecar (producer contract)"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "abandoned_mine_tunnel_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The interior of an abandoned mine tunnel featuring rugged, mineral-black rock surfaces and muted industrial textures. The environment is enveloped in deep, low-key lighting with high contrast, keeping the background mostly in heavy shadow.",
                "element_scope": "full",
                "applies_to_shots": [
                  8,
                  9
                ],
                "source_facts": [
                  "dark tunnel in restrained low-key contrast (Shot 8)",
                  "Low-key tunnel illumination (Shot 9)",
                  "mineral-black and muted industrial surfaces (Shot 9)"
                ],
                "visual_inferences": [
                  "The walls are rough and unrefined, characteristic of an old underground coal mine.",
                  "Visibility is extremely limited outside of the direct light beams, creating an oppressive atmosphere."
                ],
                "creative_decisions": [
                  "Maintaining severe high-contrast lighting across shots to emphasize the isolation and ruggedness of the underground setting."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "character_outlook",
                "id": "C01O01",
                "policy": "required"
              },
              {
                "kind": "background",
                "id": "L01B03",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "e97724b5f2749d1a"
      },
      {
        "scene_index": 1,
        "heading": "S1. INT. 아케론 딥 볼트 12, 폐광 갱도 — 밤 — 2026",
        "beat_title": "방폭문 하강",
        "representative_moment": "무거운 철제 방폭문이 갱도 천장에서 내려오며 출구를 막는 순간.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "Low angle steep tilt-up, low-key contrast lighting",
            "t2i_prompt": "Photorealistic cinematic still. Viewed from low near the ground and tilted steeply upward, a descending blast door in the middle-center midground moves toward the tunnel floor, while the tunnel floor beneath the doorway is anchored in the lower-center background. The broad iron face of the door fills the central vertical band of the L01B02 tunnel environment, suspended partway through its downward path. Its heavy lower edge hangs above the ground, defining a narrowing gap against the deep shadow of the rear opening. Low-key illumination preserves the rugged mineral face and muted industrial textures of the support frame, casting deep shadows that contrast with the severe, unyielding silhouette of the heavy metal barrier.",
            "outfit_assignments": [],
            "source_facts": [
              "S1 action: '뒤쪽 방폭문이 내려오기 시작한다.'",
              "fixed_element: 'abandoned_mine_tunnel_environment' (mineral-black rock surfaces, muted industrial textures)",
              "lighting mood: 'Low-key tunnel illumination'"
            ],
            "visual_inferences": [
              "The blast door is a heavy iron metal barrier.",
              "The descending door leaves a narrowing gap between its lower edge and the floor."
            ],
            "creative_decisions": [
              "Steep low-angle camera tilted upward to emphasize the massive weight and threat of the descending door.",
              "High contrast shadows to isolate the unyielding shape of the door against the dark tunnel background."
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001",
              "fsc_002"
            ],
            "reference_phrase_kinds": [
              "background"
            ],
            "owned_object_usage": [
              {
                "owned_token": "blast door",
                "usage_kind": "redraw",
                "source_phrase": "a descending blast door"
              },
              {
                "owned_token": "ceiling light",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "drainage channel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "dust cloud",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "gas cloud",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "inspection drone",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "mineral face",
                "usage_kind": "anchor",
                "source_phrase": "rugged mineral face"
              },
              {
                "owned_token": "mineral vein",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rail",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rock pile",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "sensor",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "support frame",
                "usage_kind": "anchor",
                "source_phrase": "support frame"
              },
              {
                "owned_token": "tunnel",
                "usage_kind": "anchor",
                "source_phrase": "tunnel floor"
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "f839fbb0d9e59eae",
              "owned_hash": "89b205ef734eeaf1",
              "camera_direction_hash": "942db074e29bcbfc",
              "owned_usage_hash": "cb9a9a838ca0a353",
              "validator": "scene_detail_owned_objects.v1",
              "violations": [
                {
                  "owned_object": "blast door",
                  "violating_phrase": "a descending blast door in the middle-center midground moves toward the tunnel floor",
                  "reason": "The dynamic descent of the blast door is explicitly required by shot_intent ('무거운 철제 방폭문이 갱도 천장에서 내려오며 출구를 막는 순간'), qualifying as a valid state change of the owned background element.",
                  "verdict": "allowed_visual_state_change"
                },
                {
                  "owned_object": "mineral face",
                  "violating_phrase": "rugged mineral face",
                  "reason": "The prompt references the mineral face purely as an existing environmental texture preserved by low-key lighting, consistent with producer's anchor declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "support frame",
                  "violating_phrase": "support frame",
                  "reason": "The prompt references the support frame only to describe its existing texture under lighting, serving as an anchor reference.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "tunnel",
                  "violating_phrase": "tunnel floor beneath the doorway is anchored in the lower-center background",
                  "reason": "The tunnel and its floor are referenced strictly as spatial framing and compositional anchors without redraw instructions.",
                  "verdict": "anchor_reference"
                }
              ]
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": 1,
        "dependency_reason": "Same room exact background consistency with Shot 8.",
        "_shot_index": 9,
        "visible_entities": [
          "L01"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 1,
            "shot_index": 9
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "wide",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "From near the tunnel floor on Tony's side of the doorway, the camera tilts steeply upward along the iron blast door's vertical path, holding it at the midpoint of its descent. The broad Tony-side face occupies the central vertical band while the descending lower edge remains above the floor, preserving the door's weight and the remaining gap beneath it.",
            "lighting_mood": "Low-key tunnel illumination preserves the mineral-black and muted industrial surfaces without softening the door's severe silhouette.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 0명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=wide — close 규칙은 이 샷에 해당하지 않는다"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "movement_direction",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "descending blast door",
                  "screen_zone": "middle_center",
                  "depth_plane": "midground",
                  "gesture_action": "moves_toward",
                  "gesture_target_label": "tunnel floor"
                },
                {
                  "constraint_id": "fsc_002",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "tunnel floor beneath doorway",
                  "screen_zone": "lower_center",
                  "depth_plane": "background",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "L01"
            ],
            "allowed_outlook_pairs": [],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "background_ref_attached",
            "bg_id": "L01B02",
            "reference_usage": "exact_background",
            "owned_objects": [
              "blast door",
              "ceiling light",
              "drainage channel",
              "dust cloud",
              "gas cloud",
              "inspection drone",
              "mineral face",
              "mineral vein",
              "rail",
              "rock pile",
              "sensor",
              "support frame",
              "tunnel"
            ],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "owned_objects lists all environment objects already drawn in the background reference — do not create new ones",
              "if focusing on an owned object, declare its reference-kind via the reference_phrase_kinds sidecar (producer contract)"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "abandoned_mine_tunnel_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The interior of an abandoned mine tunnel featuring rugged, mineral-black rock surfaces and muted industrial textures. The environment is enveloped in deep, low-key lighting with high contrast, keeping the background mostly in heavy shadow.",
                "element_scope": "full",
                "applies_to_shots": [
                  8,
                  9
                ],
                "source_facts": [
                  "dark tunnel in restrained low-key contrast (Shot 8)",
                  "Low-key tunnel illumination (Shot 9)",
                  "mineral-black and muted industrial surfaces (Shot 9)"
                ],
                "visual_inferences": [
                  "The walls are rough and unrefined, characteristic of an old underground coal mine.",
                  "Visibility is extremely limited outside of the direct light beams, creating an oppressive atmosphere."
                ],
                "creative_decisions": [
                  "Maintaining severe high-contrast lighting across shots to emphasize the isolation and ruggedness of the underground setting."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 1,
                "shot_index": 8,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "background",
                "id": "L01B02",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "e26a7f195d81fa54"
      },
      {
        "scene_index": 2,
        "heading": "S2. INT. 아케론 딥 볼트 12, 폐광 갱도 — 밤 — 2026",
        "beat_title": "갱도 붕괴와 매몰",
        "representative_moment": "굉음과 함께 갱도 천장이 무너지며 쏟아져 내리는 거대한 돌더미와 흙먼지.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "Below head height with a steep upward tilt, low-key rugged contrast",
            "t2i_prompt": "Photorealistic cinematic still. Seen from below head height with a steep upward tilt, a falling rock mass dominates the middle-center midground, moving toward the tunnel floor below the frame, while the ruptured tunnel ceiling spans the upper-center background. A dense concentration of jagged rocks descends through layered depth against the background dust cloud across the central frame. The raw, dangerous environment consists of rough, dark rock surfaces and scattered stone debris. The collapse is rendered in low-key, rugged contrast under failing illumination, casting deep shadows over the existing support frame and the fractured mineral face.",
            "outfit_assignments": [],
            "source_facts": [
              "S2 action: '갱도가 무너진다.'",
              "shot description: 굉음과 함께 갱도 천장에서 쏟아져 내리는 돌더미와 흙먼지"
            ],
            "visual_inferences": [
              "fixed_element collapsed_mine_environment: rough, dark rock surfaces and scattered stone debris",
              "low camera angle tilting up emphasizes the origin of the falling debris"
            ],
            "creative_decisions": [
              "low-key rugged contrast to highlight the raw, dangerous state of the collapse",
              "failing illumination used to cast dramatic, deep shadows across the descending rock mass"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001",
              "fsc_002"
            ],
            "reference_phrase_kinds": [
              "background"
            ],
            "owned_object_usage": [
              {
                "owned_token": "blast door",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "ceiling light",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "drainage channel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "dust cloud",
                "usage_kind": "anchor",
                "source_phrase": "against the background dust cloud"
              },
              {
                "owned_token": "gas cloud",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "inspection drone",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "mineral face",
                "usage_kind": "anchor",
                "source_phrase": "the fractured mineral face"
              },
              {
                "owned_token": "mineral vein",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rail",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rock pile",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "sensor",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "support frame",
                "usage_kind": "anchor",
                "source_phrase": "over the existing support frame"
              },
              {
                "owned_token": "tunnel",
                "usage_kind": "anchor",
                "source_phrase": "toward the tunnel floor below the frame, while the ruptured tunnel ceiling spans the upper-center background"
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "b36a46a1fb7f87ce",
              "owned_hash": "89b205ef734eeaf1",
              "camera_direction_hash": "480ae4bd11774441",
              "owned_usage_hash": "35178a3b27116829",
              "validator": "scene_detail_owned_objects.v1",
              "violations": [
                {
                  "owned_object": "dust cloud",
                  "violating_phrase": "against the background dust cloud across the central frame",
                  "reason": "The prompt references the background dust cloud as a spatial depth backdrop without requesting its recreation, matching the producer anchor declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "mineral face",
                  "violating_phrase": "the fractured mineral face",
                  "reason": "The mineral face is mentioned purely as an existing surface receiving shadows, matching the producer anchor declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "support frame",
                  "violating_phrase": "over the existing support frame",
                  "reason": "The existing support frame is referenced as an existing background anchor over which shadows are cast, matching the producer anchor declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "tunnel",
                  "violating_phrase": "toward the tunnel floor below the frame, while the ruptured tunnel ceiling spans the upper-center background",
                  "reason": "The tunnel floor and ceiling are used solely as spatial and compositional framing boundaries, matching the producer anchor declaration.",
                  "verdict": "anchor_reference"
                }
              ]
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": 1,
        "dependency_reason": "동일한 갱도 환경(L01B02)의 이전 샷 배경과의 물리적 구조 연속성 유지",
        "_shot_index": 12,
        "visible_entities": [
          "L01"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 2,
            "shot_index": 12
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "wide",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "Below head height near Tony's former position, the camera holds a steep upward tilt as the ceiling breaks apart and the densest concentration of rocks and dust crosses the central frame. Fragments descend through layered depth toward the unseen floor, with the rupturing ceiling retained along the upper frame as their point of origin.",
            "lighting_mood": "Failing tunnel illumination renders the collapse in low-key, rugged contrast as the ceiling lights begin going out in sequence.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 0명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=wide — close 규칙은 이 샷에 해당하지 않는다"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "movement_direction",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "falling rock mass",
                  "screen_zone": "middle_center",
                  "depth_plane": "midground",
                  "gesture_action": "moves_toward",
                  "gesture_target_label": "tunnel floor below frame"
                },
                {
                  "constraint_id": "fsc_002",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "ruptured tunnel ceiling",
                  "screen_zone": "upper_center",
                  "depth_plane": "background",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "L01"
            ],
            "allowed_outlook_pairs": [],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "background_ref_attached",
            "bg_id": "L01B02",
            "reference_usage": "exact_background",
            "owned_objects": [
              "blast door",
              "ceiling light",
              "drainage channel",
              "dust cloud",
              "gas cloud",
              "inspection drone",
              "mineral face",
              "mineral vein",
              "rail",
              "rock pile",
              "sensor",
              "support frame",
              "tunnel"
            ],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "owned_objects lists all environment objects already drawn in the background reference — do not create new ones",
              "if focusing on an owned object, declare its reference-kind via the reference_phrase_kinds sidecar (producer contract)"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "collapsed_mine_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The surrounding environment consists of rough, dark rock surfaces and scattered stone debris characteristic of an underground mine cave-in. Jagged rocks and loose dirt form the background and ground surfaces, exhibiting a rugged and unsettled texture.",
                "element_scope": "full",
                "applies_to_shots": [
                  12,
                  21,
                  22
                ],
                "source_facts": [
                  "abandoned mine tunnel",
                  "fragments descend through layered depth toward the unseen floor",
                  "softer tunnel background",
                  "narrow margin of surrounding debris"
                ],
                "visual_inferences": [
                  "The rock textures are dark, jagged, and heavily covered in dirt due to the recent cave-in."
                ],
                "creative_decisions": [
                  "Maintain rough and unrefined rock textures throughout the background and ground to emphasize the raw, dangerous state of the collapsed tunnel."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 1,
                "shot_index": 9,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "background",
                "id": "L01B02",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "5d38c3fe1f1174b0"
      },
      {
        "scene_index": 2,
        "heading": "S2. INT. 아케론 딥 볼트 12, 폐광 갱도 — 밤 — 2026",
        "beat_title": "물리적 시간의 정지",
        "representative_moment": "스마트폰 불빛만 남은 암흑 속에서 흙먼지 입자들이 허공에 완전히 정지해 떠 있다.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "Close framing from knee height with a diagonal tilt, shallow depth of field focusing sharply on foreground suspended dust against a soft dark background, lit by localized smartphone glow.",
            "t2i_prompt": "Photorealistic cinematic still. Seen from knee height beside a black mineral vein showing a ripple on its surface, tilted diagonally across the tunnel, completely motionless dust particles are layered from the near foreground into the middle distance. The sharply suspended dust and dirt fragments are the primary focus. The particles remain rigidly frozen against a softly out-of-focus background of rough, dark rock surfaces and scattered stone debris. The entire suspended field is lit by a weak, localized low-key glow emanating from an unseen smartphone screen out of frame, with the rest of the collapsed underground space enveloped in pitch-black heavy shadow. The solid floor is visible beneath the suspended dust, anchoring the frozen state of the air.",
            "outfit_assignments": [],
            "source_facts": [
              "completely motionless in midair",
              "abandoned mine tunnel",
              "softer tunnel background",
              "narrow margin of surrounding debris",
              "The smartphone remains the only stated illumination",
              "giving the suspended field localized low-key visibility"
            ],
            "visual_inferences": [
              "fixed_element collapsed_mine_environment: The rock textures are dark, jagged, and heavily covered in dirt due to the recent cave-in.",
              "fixed_element sole_smartphone_lighting: The light from the phone is likely a cool, bluish-white hue typical of modern digital displays, which contrasts sharply with the surrounding dark environment."
            ],
            "creative_decisions": [
              "Maintain rough and unrefined rock textures throughout the background to emphasize the raw state of the collapsed tunnel.",
              "Use extremely high-contrast, localized lighting to heighten the sense of isolation and focus visual attention entirely on the suspended dust.",
              "Focus purely on the suspended particles and the black vein to visually convey the unnatural arrest of time, keeping the character out of frame."
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [],
            "reference_phrase_kinds": [],
            "owned_object_usage": [
              {
                "owned_token": "blast door",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "ceiling",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "ceiling light",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "concrete crack",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "concrete lining",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "drainage channel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "dust",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "inspection drone",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "mineral vein",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rail",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rock pile",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "sensor",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "smartphone",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "support frame",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tree root",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tunnel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "water droplet",
                "usage_kind": "absent",
                "source_phrase": ""
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "3247a1d3b5dc7f66",
              "owned_hash": "51e804249a9c8214",
              "camera_direction_hash": "f48b2dd372371ddd",
              "owned_usage_hash": "e18b11a66d536d4f",
              "validator": "scene_detail_owned_objects.v1.close_skip",
              "violations": []
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": 2,
        "dependency_reason": "같은 폐광 공간의 붕괴 직후 상황 유지",
        "_shot_index": 21,
        "visible_entities": [],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 2,
            "shot_index": 21
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "close",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "At knee height beside the black vein, the tilt settles into a near-static diagonal view across the tunnel, with motionless dust particles layered from the near foreground into the middle distance. The particles remain sharply suspended against a softer tunnel background, while Tony stays outside the crop in the direction of the camera's impending advance.",
            "lighting_mood": "The smartphone remains the only stated illumination, giving the suspended field localized low-key visibility without introducing another light source.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 0명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "primary_subject_only_fully_visible": "the named primary subject (hand / face / object) is the only fully visible entity in the frame",
                  "other_entities_appearance_options": [
                    "partial visibility — shoulder slice / arm slice / sleeve / fingertips entering the frame edge",
                    "soft background — out-of-focus silhouette via shallow depth of field",
                    "absent — not rendered at all even if listed in visible_entities"
                  ],
                  "forbidden_combinations": [
                    "same character's full body and body-part close-up in the same frame (view mixing)",
                    "two characters both rendered as sharp close-up faces in the same frame"
                  ],
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "wide_medium_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=close — wide/medium 규칙은 이 샷에 해당하지 않는다"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": null,
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [],
            "allowed_outlook_pairs": [],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "skipped_close_framing",
            "bg_id": "L01B01",
            "reference_usage": "skipped_close_framing",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": true,
            "constraints": [
              "this shot uses close framing — the background reference is intentionally omitted at composition time",
              "describe the close subject and its immediate surroundings without relying on the absent background reference",
              "use [L##: ...] block for frame-edge surface description"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "collapsed_mine_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The surrounding environment consists of rough, dark rock surfaces and scattered stone debris characteristic of an underground mine cave-in. Jagged rocks and loose dirt form the background and ground surfaces, exhibiting a rugged and unsettled texture.",
                "element_scope": "full",
                "applies_to_shots": [
                  12,
                  21,
                  22
                ],
                "source_facts": [
                  "abandoned mine tunnel",
                  "fragments descend through layered depth toward the unseen floor",
                  "softer tunnel background",
                  "narrow margin of surrounding debris"
                ],
                "visual_inferences": [
                  "The rock textures are dark, jagged, and heavily covered in dirt due to the recent cave-in."
                ],
                "creative_decisions": [
                  "Maintain rough and unrefined rock textures throughout the background and ground to emphasize the raw, dangerous state of the collapsed tunnel."
                ],
                "confidence": "high"
              },
              {
                "element_id": "sole_smartphone_lighting",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The underground space is enveloped in pitch-black darkness due to the failure of all ceiling lights. The only active illumination is a weak, localized glow emanating from a smartphone screen, which casts low-key light on the immediate surrounding area while the rest of the space remains in deep, heavy shadow.",
                "element_scope": "full",
                "applies_to_shots": [
                  21,
                  22
                ],
                "source_facts": [
                  "The smartphone remains the only stated illumination",
                  "giving the suspended field localized low-key visibility",
                  "The smartphone display provides the only stated localized light",
                  "isolated within otherwise low-key darkness"
                ],
                "visual_inferences": [
                  "The light from the phone is likely a cool, bluish-white hue typical of modern digital displays, which contrasts sharply with the surrounding dark environment."
                ],
                "creative_decisions": [
                  "Use extremely high-contrast, localized lighting to heighten the sense of isolation and to focus visual attention entirely on the small area reached by the screen's glow."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 2,
                "shot_index": 12,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [],
            "forbidden_refs": [
              {
                "kind": "background",
                "reason": "close framing skips chain_bg reference"
              }
            ],
            "readiness_policy": "skipped_by_policy",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": true,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "936d929562856da5"
      },
      {
        "scene_index": 2,
        "heading": "S2. INT. 아케론 딥 볼트 12, 폐광 갱도 — 밤 — 2026",
        "beat_title": "시간의 정지",
        "representative_moment": "A macro insert of the P01 smartphone screen glowing in the dark, its clock digits frozen at 21:14:07.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "steep oblique overhead insert shot with localized bluish screen glow",
            "t2i_prompt": "Photorealistic cinematic still. Seen from a steep oblique overhead angle, the P01 cracked 2020s smartphone rests on the rugged surface, its active smartphone screen showing 21:14:07 holding the central visual weight in the middle-center foreground. The device is enveloped in pitch-black darkness, providing the only illumination via a weak, localized bluish glow emanating from the display. The screen's bezel and cracked glass are clearly defined. Around the device, a narrow margin of the existing dark rock pile and loose ground of the background is faintly caught in the low-key screen light, emphasizing the frozen instant in the abandoned mine tunnel.",
            "outfit_assignments": [],
            "source_facts": [
              "스마트폰 시계가 21:14:07에서 멈춘다.",
              "스마트폰 불빛만 남는다.",
              "주변 먼지가 공중에서 멈춘다."
            ],
            "visual_inferences": [
              "fixed_element collapsed_mine_environment: rough, dark rock surfaces and jagged rock pile",
              "fixed_element sole_smartphone_lighting: pitch-black darkness with weak, localized glow from the screen",
              "steep oblique overhead angle makes the screen readable while maintaining the device's physical presence"
            ],
            "creative_decisions": [
              "cool bluish digital glow contrasting with the deep shadows of the rock debris",
              "shallow depth of field focused entirely on the glowing digits to anchor the frozen temporal state"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001"
            ],
            "reference_phrase_kinds": [
              "background"
            ],
            "owned_object_usage": [
              {
                "owned_token": "blast door",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "ceiling",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "ceiling light",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "concrete crack",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "concrete lining",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "drainage channel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "dust",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "inspection drone",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "mineral vein",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rail",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rock pile",
                "usage_kind": "anchor",
                "source_phrase": "existing dark rock pile and loose ground of the background"
              },
              {
                "owned_token": "sensor",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "smartphone",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "support frame",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tree root",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tunnel",
                "usage_kind": "anchor",
                "source_phrase": "in the abandoned mine tunnel"
              },
              {
                "owned_token": "water droplet",
                "usage_kind": "absent",
                "source_phrase": ""
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "45ac86b36b14ff24",
              "owned_hash": "51e804249a9c8214",
              "camera_direction_hash": "b2302b52a9eea0ff",
              "owned_usage_hash": "59f9d9ded04a6805",
              "validator": "scene_detail_owned_objects.v1",
              "violations": [
                {
                  "owned_object": "rock pile",
                  "violating_phrase": "existing dark rock pile and loose ground of the background",
                  "reason": "The prompt references the existing rock pile purely as ambient background caught in the screen light, matching the producer's anchor declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "smartphone",
                  "violating_phrase": "the P01 cracked 2020s smartphone rests on the rugged surface, its active smartphone screen showing 21:14:07",
                  "reason": "Producer declared absent despite prominent appearance, but the screen state and glow are explicitly and structurally required by shot_intent ('A macro insert of the P01 smartphone screen glowing in the dark, its clock digits frozen at 21:14:07') satisfying narrow exception (1).",
                  "verdict": "allowed_visual_state_change"
                },
                {
                  "owned_object": "tunnel",
                  "violating_phrase": "in the abandoned mine tunnel",
                  "reason": "The tunnel is mentioned solely as a background environmental setting anchor, matching the producer's anchor declaration.",
                  "verdict": "anchor_reference"
                }
              ]
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": 2,
        "dependency_reason": "Shot 21에서 확립된 바닥의 돌더미(collapsed_mine_environment)와 유일한 광원(sole_smartphone_lighting)의 시각적 연속성 유지",
        "_shot_index": 22,
        "visible_entities": [
          "P01"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 2,
            "shot_index": 22
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "insert",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "The push-in completes above and to one side of the smartphone at a steep oblique overhead angle, keeping the screen plane readable while its bezel and a narrow margin of surrounding debris preserve the device as a physical object. The clock digits 21:14:07 hold the central visual weight at the exact instant they cease changing, with the phone kept within a natural proportion to its surroundings.",
            "lighting_mood": "The smartphone display provides the only stated localized light, isolated within otherwise low-key darkness.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 0명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "primary_subject_only_fully_visible": "the named primary subject (hand / face / object) is the only fully visible entity in the frame",
                  "other_entities_appearance_options": [
                    "partial visibility — shoulder slice / arm slice / sleeve / fingertips entering the frame edge",
                    "soft background — out-of-focus silhouette via shallow depth of field",
                    "absent — not rendered at all even if listed in visible_entities"
                  ],
                  "forbidden_combinations": [
                    "same character's full body and body-part close-up in the same frame (view mixing)",
                    "two characters both rendered as sharp close-up faces in the same frame"
                  ],
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "required_background_position",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "smartphone screen showing 21:14:07",
                  "screen_zone": "middle_center",
                  "depth_plane": "foreground",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "P01"
            ],
            "allowed_outlook_pairs": [],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": true,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "background_ref_attached",
            "bg_id": "L01B01",
            "reference_usage": "exact_background",
            "owned_objects": [
              "blast door",
              "ceiling",
              "ceiling light",
              "concrete crack",
              "concrete lining",
              "drainage channel",
              "dust",
              "inspection drone",
              "mineral vein",
              "rail",
              "rock pile",
              "sensor",
              "smartphone",
              "support frame",
              "tree root",
              "tunnel",
              "water droplet"
            ],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "owned_objects lists all environment objects already drawn in the background reference — do not create new ones",
              "if focusing on an owned object, declare its reference-kind via the reference_phrase_kinds sidecar (producer contract)"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "collapsed_mine_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The surrounding environment consists of rough, dark rock surfaces and scattered stone debris characteristic of an underground mine cave-in. Jagged rocks and loose dirt form the background and ground surfaces, exhibiting a rugged and unsettled texture.",
                "element_scope": "full",
                "applies_to_shots": [
                  12,
                  21,
                  22
                ],
                "source_facts": [
                  "abandoned mine tunnel",
                  "fragments descend through layered depth toward the unseen floor",
                  "softer tunnel background",
                  "narrow margin of surrounding debris"
                ],
                "visual_inferences": [
                  "The rock textures are dark, jagged, and heavily covered in dirt due to the recent cave-in."
                ],
                "creative_decisions": [
                  "Maintain rough and unrefined rock textures throughout the background and ground to emphasize the raw, dangerous state of the collapsed tunnel."
                ],
                "confidence": "high"
              },
              {
                "element_id": "sole_smartphone_lighting",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The underground space is enveloped in pitch-black darkness due to the failure of all ceiling lights. The only active illumination is a weak, localized glow emanating from a smartphone screen, which casts low-key light on the immediate surrounding area while the rest of the space remains in deep, heavy shadow.",
                "element_scope": "full",
                "applies_to_shots": [
                  21,
                  22
                ],
                "source_facts": [
                  "The smartphone remains the only stated illumination",
                  "giving the suspended field localized low-key visibility",
                  "The smartphone display provides the only stated localized light",
                  "isolated within otherwise low-key darkness"
                ],
                "visual_inferences": [
                  "The light from the phone is likely a cool, bluish-white hue typical of modern digital displays, which contrasts sharply with the surrounding dark environment."
                ],
                "creative_decisions": [
                  "Use extremely high-contrast, localized lighting to heighten the sense of isolation and to focus visual attention entirely on the small area reached by the screen's glow."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 2,
                "shot_index": 21,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "background",
                "id": "L01B01",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "14c47268f9d44c49"
      },
      {
        "scene_index": 3,
        "heading": "INT. 아케론 딥 볼트 12, 폐광 갱도 — 밤 — 2026",
        "beat_title": "시간이 멈춘 갱도 내 정지된 물방울",
        "representative_moment": "C01의 굳어 있는 얼굴 옆 허공에 물방울 하나가 얼어붙듯 정지해 있다.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "cheek-height close framing with a static three-quarter profile and restrained low-key illumination",
            "t2i_prompt": "Photorealistic cinematic still. C01O01, an American man in his mid-30s in a black disaster rescue workwear, is visible in the middle-left midground presenting a rigid three-quarter left profile, alongside a suspended water droplet hanging motionless in the middle-center foreground. Seen from cheek height and very close, the figure occupies the left side from the neck up. His face and neck are rigid, eyes fixed off-screen. The droplet rests in the adjacent focal plane, separated from the profile by restrained low-key illumination. Motionless dust hangs suspended in the air. The immediate surroundings dissolve into the deep shadows and weathered debris of the enclosed tunnel backdrop, maintaining a dark and moody atmosphere.",
            "outfit_assignments": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "source_facts": [
              "S3 action: '균열에서 물 한 방울이 솟지만, 토니 가까이에서 얼어붙듯 멎는다.'",
              "shot description: 굳어 있는 토니의 얼굴 옆 허공에 얼어붙은 듯 맺혀 있는 물 한 방울",
              "camera_direction: At cheek height and very close to 토니, static three-quarter profile, suspended water droplet in adjacent focal plane"
            ],
            "visual_inferences": [
              "fixed_element: The enclosed, collapsed mine tunnel serves as a constant backdrop featuring weathered debris",
              "fixed_element: Restrained low-key illumination casts deep shadows across the ruins",
              "close framing removes broad room geometry, leaving only the neck-up subject and adjacent droplet visible"
            ],
            "creative_decisions": [
              "cheek-height camera matches the vertical alignment of the head and the droplet",
              "deep shadow at the frame edges isolates the primary subjects, enhancing the surreal stillness of the suspended water and dust"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001",
              "fsc_002"
            ],
            "reference_phrase_kinds": [
              "character"
            ],
            "owned_object_usage": [
              {
                "owned_token": "blast door",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "ceiling",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "ceiling light",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "concrete crack",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "concrete lining",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "drainage channel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "dust",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "inspection drone",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "mineral vein",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rail",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rock pile",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "sensor",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "smartphone",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "support frame",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tree root",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tunnel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "water droplet",
                "usage_kind": "absent",
                "source_phrase": ""
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "02e39513124ec232",
              "owned_hash": "51e804249a9c8214",
              "camera_direction_hash": "4f3b4e70bafcffc9",
              "owned_usage_hash": "e18b11a66d536d4f",
              "validator": "scene_detail_owned_objects.v1.close_skip",
              "violations": []
            }
          }
        ],
        "scene_type": "montage",
        "dependent_scene_index": 2,
        "dependency_reason": "S2_Shot21의 공간적 일관성 유지 (close framing 규칙에 따라 인접한 환경 및 조명 톤만 유지)",
        "_shot_index": 26,
        "visible_entities": [
          "C01"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 3,
            "shot_index": 26
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "close",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "At cheek height and very close to 토니(앤서니 로저스), a static three-quarter profile composition places his rigid face along the left plane and the suspended water droplet just right of center in the adjacent focal plane. The droplet occupies the same match-cut region as the preceding suspended detail, while Tony's fixed eyes remain directed beyond the frame rather than toward the lens.",
            "lighting_mood": "Restrained low-key illumination appropriate to the enclosed tunnel separates the droplet and Tony's rigid profile without specifying an additional source.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 1명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "primary_subject_only_fully_visible": "the named primary subject (hand / face / object) is the only fully visible entity in the frame",
                  "other_entities_appearance_options": [
                    "partial visibility — shoulder slice / arm slice / sleeve / fingertips entering the frame edge",
                    "soft background — out-of-focus silhouette via shallow depth of field",
                    "absent — not rendered at all even if listed in visible_entities"
                  ],
                  "forbidden_combinations": [
                    "same character's full body and body-part close-up in the same frame (view mixing)",
                    "two characters both rendered as sharp close-up faces in the same frame"
                  ],
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "wide_medium_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=close — wide/medium 규칙은 이 샷에 해당하지 않는다"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "shared_space_relation",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "suspended water droplet",
                  "screen_zone": "middle_center",
                  "depth_plane": "foreground",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                },
                {
                  "constraint_id": "fsc_002",
                  "target_kind": "character",
                  "target_id": "C01",
                  "label": "토니(앤서니 로저스) rigid profile",
                  "screen_zone": "middle_left",
                  "depth_plane": "midground",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "C01"
            ],
            "allowed_outlook_pairs": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "skipped_close_framing",
            "bg_id": "L01B01",
            "reference_usage": "skipped_close_framing",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": true,
            "constraints": [
              "this shot uses close framing — the background reference is intentionally omitted at composition time",
              "describe the close subject and its immediate surroundings without relying on the absent background reference",
              "use [L##: ...] block for frame-edge surface description"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "enclosed_tunnel_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The enclosed, collapsed mine tunnel serves as a constant backdrop, featuring weathered debris and signs of natural decay over time. Restrained low-key illumination casts deep shadows across the ruins, maintaining a dark and moody atmosphere without introducing any new or explicitly visible light sources.",
                "element_scope": "full",
                "applies_to_shots": [
                  26,
                  30,
                  31
                ],
                "source_facts": [
                  "Restrained low-key illumination appropriate to the enclosed tunnel",
                  "weathered remnants occupy more of the frame",
                  "붕괴된 갱도 잔해 상태",
                  "자연에 잠식된 상태"
                ],
                "visual_inferences": [
                  "Deep shadows and low visibility dominate the background due to the minimal lighting setup"
                ],
                "creative_decisions": [
                  "Using a unified dark, moody lighting scheme across the match-cut sequence to emphasize the passage of time without altering the spatial lighting"
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 2,
                "shot_index": 21,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "character_outlook",
                "id": "C01O01",
                "policy": "required"
              }
            ],
            "forbidden_refs": [
              {
                "kind": "background",
                "reason": "close framing skips chain_bg reference"
              }
            ],
            "readiness_policy": "skipped_by_policy",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": true,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "b47864701f8fcb13"
      },
      {
        "scene_index": 3,
        "heading": "S3. INT. 아케론 딥 볼트 12, 폐광 갱도 — 밤 — 2026",
        "beat_title": "뿌리에 갈라지는 콘크리트 벽",
        "representative_moment": "굵어진 뿌리의 팽창 압력을 이기지 못하고 쩍 갈라진 콘크리트 벽면.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "close wall-side observer angle at waist height, restrained low-key illumination",
            "t2i_prompt": "Photorealistic cinematic still. Seen from a close wall-side position at waist height and off-axis, a thick root and a ruptured concrete wall dominate the frame. The root's pressure point sits low on one side, where the root, expanded to rope-like thickness, presses outward. Its side contour crosses the pressure point diagonally before continuing beyond the frame edge. The wall is split open under the pressure, an opened fracture running diagonally across the concrete surface. Broken concrete fragments are dislodged around the split. Restrained low-key illumination emphasizes the rough texture of the wall break and root contour, casting deep shadows across the [L01: enclosed mine tunnel ruins and weathered debris at the frame edges], maintaining a dark and moody atmosphere without explicitly visible light sources.",
            "outfit_assignments": [],
            "source_facts": [
              "S3 action: '콘크리트가 뿌리 압력에 갈라진다.'",
              "shot description: concrete wall split open under root pressure, thick root expanded to rope-like thickness",
              "camera_direction: From a close wall-side position at waist height, off-axis, the thick root's pressure point sits low on one side and the opened fracture runs diagonally across the concrete wall"
            ],
            "visual_inferences": [
              "fixed_element enclosed_tunnel_environment: Restrained low-key illumination casts deep shadows across the ruins and maintains a dark and moody atmosphere"
            ],
            "creative_decisions": [
              "translated DP's close-up requirement into 'dominate the frame' and 'frame edge' to describe the exact visual boundaries without relying on camera taxonomy",
              "emphasizing the rough texture of the broken concrete and the massive root under low-key illumination to highlight the profound passage of time"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [],
            "reference_phrase_kinds": [],
            "owned_object_usage": [
              {
                "owned_token": "blast door",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "ceiling",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "ceiling light",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "concrete crack",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "concrete lining",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "drainage channel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "dust",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "inspection drone",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "mineral vein",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rail",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rock pile",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "sensor",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "smartphone",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "support frame",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tree root",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tunnel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "water droplet",
                "usage_kind": "absent",
                "source_phrase": ""
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "d2116542cf1b40f6",
              "owned_hash": "51e804249a9c8214",
              "camera_direction_hash": "3396f02af6cbd603",
              "owned_usage_hash": "e18b11a66d536d4f",
              "validator": "scene_detail_owned_objects.v1.close_skip",
              "violations": []
            }
          }
        ],
        "scene_type": "montage",
        "dependent_scene_index": 2,
        "dependency_reason": "S2의 갱도 내부 배경(L01)과 조명, 분위기 설정(enclosed_tunnel_environment)을 그대로 유지하며 자연에 의한 파괴 묘사",
        "_shot_index": 30,
        "visible_entities": [
          "L01"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 3,
            "shot_index": 30
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "close",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "From a close wall-side position at waist height, the camera remains static and off-axis as the thick root's pressure point sits low on one side and the opened fracture runs diagonally across the concrete wall. The frame holds until the split is fully spread, preserving the rupture as the single changed visual axis within the match-cut sequence.",
            "lighting_mood": "Restrained low-key illumination emphasizes the wall break and root contour without introducing a new visible light source.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 0명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "primary_subject_only_fully_visible": "the named primary subject (hand / face / object) is the only fully visible entity in the frame",
                  "other_entities_appearance_options": [
                    "partial visibility — shoulder slice / arm slice / sleeve / fingertips entering the frame edge",
                    "soft background — out-of-focus silhouette via shallow depth of field",
                    "absent — not rendered at all even if listed in visible_entities"
                  ],
                  "forbidden_combinations": [
                    "same character's full body and body-part close-up in the same frame (view mixing)",
                    "two characters both rendered as sharp close-up faces in the same frame"
                  ],
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "wide_medium_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=close — wide/medium 규칙은 이 샷에 해당하지 않는다"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": null,
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "L01"
            ],
            "allowed_outlook_pairs": [],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "skipped_close_framing",
            "bg_id": "L01B01",
            "reference_usage": "skipped_close_framing",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": true,
            "constraints": [
              "this shot uses close framing — the background reference is intentionally omitted at composition time",
              "describe the close subject and its immediate surroundings without relying on the absent background reference",
              "use [L##: ...] block for frame-edge surface description"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "enclosed_tunnel_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The enclosed, collapsed mine tunnel serves as a constant backdrop, featuring weathered debris and signs of natural decay over time. Restrained low-key illumination casts deep shadows across the ruins, maintaining a dark and moody atmosphere without introducing any new or explicitly visible light sources.",
                "element_scope": "full",
                "applies_to_shots": [
                  26,
                  30,
                  31
                ],
                "source_facts": [
                  "Restrained low-key illumination appropriate to the enclosed tunnel",
                  "weathered remnants occupy more of the frame",
                  "붕괴된 갱도 잔해 상태",
                  "자연에 잠식된 상태"
                ],
                "visual_inferences": [
                  "Deep shadows and low visibility dominate the background due to the minimal lighting setup"
                ],
                "creative_decisions": [
                  "Using a unified dark, moody lighting scheme across the match-cut sequence to emphasize the passage of time without altering the spatial lighting"
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 2,
                "shot_index": 12,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [],
            "forbidden_refs": [
              {
                "kind": "background",
                "reason": "close framing skips chain_bg reference"
              }
            ],
            "readiness_policy": "skipped_by_policy",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": true,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "3ed45c0917345dfa"
      },
      {
        "scene_index": 3,
        "heading": "INT. 아케론 딥 볼트 12, 폐광 갱도 — 밤 — 2026",
        "beat_title": "시간이 멈춘 스마트폰과 풍화되는 갱도",
        "representative_moment": "가파른 하이앵글 아래, P01이 부서진 잔해 위에서 여전히 '21:14:07'을 띄우고 있다.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "Steep high-angle close framing with neutral low-key illumination",
            "t2i_prompt": "Photorealistic cinematic still. Seen from a steep high angle, P01, a cracked smartphone screen showing 21:14:07, rests in the lower-center foreground, while weathered debris surrounding phone dominates the middle-center midground of the [L01: dark mine tunnel floor]. Restrained neutral low-key illumination casts deep shadows across the close-up ruins, keeping the glowing digital interface clearly legible against the natural decay. The camera focuses sharply on the shattered glass of the device, blurring the encroaching roots and rubble at the frame edges, maintaining a dark and moody atmosphere.",
            "outfit_assignments": [],
            "source_facts": [
              "스마트폰 시계는 계속 21:14:07",
              "weathered remnants occupy more of the frame",
              "Steep high-angle view of the smartphone among the debris"
            ],
            "visual_inferences": [
              "fixed_element enclosed_tunnel_environment: Restrained low-key illumination casts deep shadows across the ruins",
              "Camera focuses sharply on the phone screen with shallow depth of field for the surrounding debris"
            ],
            "creative_decisions": [
              "Placed P01 in the lower-center foreground and debris in the midground to strictly follow the frame spatial contract",
              "Emphasized the glow of the screen against the dark, moody low-key lighting of the tunnel floor"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001",
              "fsc_002"
            ],
            "reference_phrase_kinds": [],
            "owned_object_usage": [
              {
                "owned_token": "blast door",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "ceiling",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "ceiling light",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "concrete crack",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "concrete lining",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "drainage channel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "dust",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "inspection drone",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "mineral vein",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rail",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rock pile",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "sensor",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "smartphone",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "support frame",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tree root",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tunnel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "water droplet",
                "usage_kind": "absent",
                "source_phrase": ""
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "edca562555f65bb0",
              "owned_hash": "51e804249a9c8214",
              "camera_direction_hash": "c376710de24293a6",
              "owned_usage_hash": "e18b11a66d536d4f",
              "validator": "scene_detail_owned_objects.v1.close_skip",
              "violations": []
            }
          }
        ],
        "scene_type": "montage",
        "dependent_scene_index": 3,
        "dependency_reason": "몽타주 시퀀스 내에서 이전 샷(Shot 30)의 고정된 환경 톤(enclosed_tunnel_environment)과 조명을 그대로 이어받음",
        "_shot_index": 31,
        "visible_entities": [
          "L01",
          "P01"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 3,
            "shot_index": 31
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "close",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "From the same wall-side base position, the downward tilt settles into a steep high-angle view of the smartphone among the debris. The screen sits near the lower center and remains clearly legible at 21:14:07, while the surrounding weathered remnants occupy more of the frame than the device and register the elapsed environmental change.",
            "lighting_mood": "Neutral low-key illumination appropriate to the tunnel keeps the screen legible against the restrained, weathered surroundings without specifying a new source.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 0명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "primary_subject_only_fully_visible": "the named primary subject (hand / face / object) is the only fully visible entity in the frame",
                  "other_entities_appearance_options": [
                    "partial visibility — shoulder slice / arm slice / sleeve / fingertips entering the frame edge",
                    "soft background — out-of-focus silhouette via shallow depth of field",
                    "absent — not rendered at all even if listed in visible_entities"
                  ],
                  "forbidden_combinations": [
                    "same character's full body and body-part close-up in the same frame (view mixing)",
                    "two characters both rendered as sharp close-up faces in the same frame"
                  ],
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "wide_medium_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=close — wide/medium 규칙은 이 샷에 해당하지 않는다"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "required_background_position",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "smartphone screen showing 21:14:07",
                  "screen_zone": "lower_center",
                  "depth_plane": "foreground",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                },
                {
                  "constraint_id": "fsc_002",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "weathered debris surrounding phone",
                  "screen_zone": "middle_center",
                  "depth_plane": "midground",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "L01",
              "P01"
            ],
            "allowed_outlook_pairs": [],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": true,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "skipped_close_framing",
            "bg_id": "L01B01",
            "reference_usage": "skipped_close_framing",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": true,
            "constraints": [
              "this shot uses close framing — the background reference is intentionally omitted at composition time",
              "describe the close subject and its immediate surroundings without relying on the absent background reference",
              "use [L##: ...] block for frame-edge surface description"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "enclosed_tunnel_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The enclosed, collapsed mine tunnel serves as a constant backdrop, featuring weathered debris and signs of natural decay over time. Restrained low-key illumination casts deep shadows across the ruins, maintaining a dark and moody atmosphere without introducing any new or explicitly visible light sources.",
                "element_scope": "full",
                "applies_to_shots": [
                  26,
                  30,
                  31
                ],
                "source_facts": [
                  "Restrained low-key illumination appropriate to the enclosed tunnel",
                  "weathered remnants occupy more of the frame",
                  "붕괴된 갱도 잔해 상태",
                  "자연에 잠식된 상태"
                ],
                "visual_inferences": [
                  "Deep shadows and low visibility dominate the background due to the minimal lighting setup"
                ],
                "creative_decisions": [
                  "Using a unified dark, moody lighting scheme across the match-cut sequence to emphasize the passage of time without altering the spatial lighting"
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 3,
                "shot_index": 30,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [],
            "forbidden_refs": [
              {
                "kind": "background",
                "reason": "close framing skips chain_bg reference"
              }
            ],
            "readiness_policy": "skipped_by_policy",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": true,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "eaf862dddb7448b9"
      },
      {
        "scene_index": 4,
        "heading": "S4. INT. 같은 갱도 — 알 수 없는 시간",
        "beat_title": "토니의 각성과 이질적인 갱도 환경",
        "representative_moment": "막혔던 숨을 토해내며 눈을 뜬 토니의 얼굴 줌인.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "Intimate close framing from slightly above, shallow depth of field, restrained low-key lighting",
            "t2i_prompt": "Photorealistic cinematic still. C01O01 in a dark workwear jacket, a Caucasian man in his 30s, his breathing face occupies the middle-center foreground. Captured at an intimate close range from slightly above, the camera angles down at a three-quarter angle from the right, keeping his open mouth and eyes together in sharp focus. His dust-marked face fills most of the frame, with a slice of his shoulders extending toward the lower frame edge. His chin is lifted and his neck is tensed, abruptly releasing a breath, looking off-screen into the nearby darkness rather than addressing the lens. Restrained low-key ambient illumination emphasizes the tactile dust on his skin while preserving deep shadows that wrap around the periphery. The out-of-focus background consists of weathered neutral tones and mineral blacks, enveloping the close subject in a somber, uncertain mood without revealing specific structures.",
            "outfit_assignments": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "source_facts": [
              "S4 text: 토니가 갑자기 숨을 들이마신다.",
              "Shot 32 description: 막혔던 숨을 토해내며 입을 크게 벌린 토니의 흙먼지 묻은 얼굴.",
              "Character angles/state: three_quarter_right, chin lifted, mouth open, gaze=off_screen, state=alive"
            ],
            "visual_inferences": [
              "fixed_element tunnel_low_key_lighting: Restrained low-key illumination dominates the space, preserving deep shadows, mineral blacks, and weathered neutral tones.",
              "Dust on the face made tactile by the restrained ambient light.",
              "The background is rendered as a soft, out-of-focus void to comply with skipped_close_framing, omitting the absent reference."
            ],
            "creative_decisions": [
              "Intimate close framing from slightly above, ensuring the shoulders extend to the lower frame edge for geometric consistency.",
              "Shallow depth of field to isolate the tense facial expression against the deep shadows of the environment."
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001"
            ],
            "reference_phrase_kinds": [
              "character"
            ],
            "owned_object_usage": [
              {
                "owned_token": "blast door",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "light",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "mine portal",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "root",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "sensor",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "threshold",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tunnel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "wall",
                "usage_kind": "absent",
                "source_phrase": ""
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "ad5ddc3c8e3a552b",
              "owned_hash": "3647dc307f207768",
              "camera_direction_hash": "35dedec393d7bf32",
              "owned_usage_hash": "d0b1729fb7b22497",
              "validator": "scene_detail_owned_objects.v1.close_skip",
              "violations": []
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": -1,
        "dependency_reason": "N/A",
        "_shot_index": 32,
        "visible_entities": [
          "C01"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 4,
            "shot_index": 32
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "close",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "At intimate range slightly above 토니(앤서니 로저스), the handheld camera angles down from his front three-quarter side so his dust-marked face occupies most of the frame while his open mouth and eyes remain together in focus. His chin lifts and neck tenses as he abruptly releases the breath, eyes searching into the nearby off-screen tunnel rather than addressing the lens.",
            "lighting_mood": "Restrained low-key illumination keeps the dust on Tony's face tactile while preserving the dark, uncertain tunnel mood without naming an unstated source.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 1명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "primary_subject_only_fully_visible": "the named primary subject (hand / face / object) is the only fully visible entity in the frame",
                  "other_entities_appearance_options": [
                    "partial visibility — shoulder slice / arm slice / sleeve / fingertips entering the frame edge",
                    "soft background — out-of-focus silhouette via shallow depth of field",
                    "absent — not rendered at all even if listed in visible_entities"
                  ],
                  "forbidden_combinations": [
                    "same character's full body and body-part close-up in the same frame (view mixing)",
                    "two characters both rendered as sharp close-up faces in the same frame"
                  ],
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "wide_medium_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=close — wide/medium 규칙은 이 샷에 해당하지 않는다"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "primary_subject_isolation",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "character",
                  "target_id": "C01",
                  "label": "토니(앤서니 로저스) breathing face",
                  "screen_zone": "middle_center",
                  "depth_plane": "foreground",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "C01"
            ],
            "allowed_outlook_pairs": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "skipped_close_framing",
            "bg_id": "L02B01",
            "reference_usage": "skipped_close_framing",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": true,
            "constraints": [
              "this shot uses close framing — the background reference is intentionally omitted at composition time",
              "describe the close subject and its immediate surroundings without relying on the absent background reference",
              "use [L##: ...] block for frame-edge surface description"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "tunnel_low_key_lighting",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The surrounding environment is a dark, ruined underground tunnel. Restrained low-key illumination dominates the space, preserving deep shadows, mineral blacks, and weathered neutral tones. This consistent lighting condition creates a somber, uncertain, and oppressive mood throughout the area, enveloping all subjects and objects.",
                "element_scope": "full",
                "applies_to_shots": [
                  32,
                  33,
                  36
                ],
                "source_facts": [
                  "preserving the dark, uncertain tunnel mood without naming an unstated source",
                  "an otherwise restrained low-key frame",
                  "Low-to-mid-key ambient light appropriate to the tunnel preserves restrained mineral blacks and weathered neutral tones"
                ],
                "visual_inferences": [
                  "The lack of a prominent light source means deep shadows naturally wrap around the periphery of all elements, seamlessly tying the separate shots into a single dark location."
                ],
                "creative_decisions": [
                  "Maintain a unified low-key exposure and shadow depth across different subjects (the face, the phone screen, the roots) to firmly anchor them in the same collapsed tunnel environment."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "character_outlook",
                "id": "C01O01",
                "policy": "required"
              }
            ],
            "forbidden_refs": [
              {
                "kind": "background",
                "reason": "close framing skips chain_bg reference"
              }
            ],
            "readiness_policy": "skipped_by_policy",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": true,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "9d85002d7f43b0d5"
      },
      {
        "scene_index": 4,
        "heading": "INT. 같은 갱도 — 알 수 없는 시간 (01:50–02:35)",
        "beat_title": "시간의 재개",
        "representative_moment": "스마트폰 시계가 21:14:08로 바뀌며 시간이 다시 흐르기 시작하는 순간.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "overhead direct insert shot, screen-aligned, localized screen illumination",
            "t2i_prompt": "Photorealistic cinematic still. P01 rests flat in the middle-center foreground, the smartphone display showing 21:14:08 and 17 percent on its cracked touchscreen. Directly above the device and aligned with its screen plane, the camera looks straight down. The screen sits centrally, retaining its black bezel and a small surrounding margin of the weathered tunnel floor. The active display supplies localized screen illumination, while an otherwise restrained low-key frame and deep shadows dominate the periphery, preserving mineral blacks and a somber, uncertain mood.",
            "outfit_assignments": [],
            "source_facts": [
              "S4 action: '스마트폰 시계가 21:14:08로 넘어간다. 배터리 17퍼센트.'",
              "shot description: 21:14:08'로 숫자가 넘어간 순간의 스마트폰 화면 클로즈업."
            ],
            "visual_inferences": [
              "fixed_element tunnel_low_key_lighting: low-key exposure preserving deep shadows and mineral blacks",
              "the phone rests flat on the ground to allow an overhead screen-aligned shot"
            ],
            "creative_decisions": [
              "tight framing keeping only a small margin of the tunnel floor to maintain spatial context without diluting the insert framing",
              "screen glow contrasting with the deep shadows of the tunnel environment to emphasize the resumption of time"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001"
            ],
            "reference_phrase_kinds": [
              "background"
            ],
            "owned_object_usage": [
              {
                "owned_token": "blast door",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "light",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "mine portal",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "root",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "sensor",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "threshold",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tunnel",
                "usage_kind": "anchor",
                "source_phrase": "weathered tunnel floor"
              },
              {
                "owned_token": "wall",
                "usage_kind": "absent",
                "source_phrase": ""
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "e7964f6bbee3006a",
              "owned_hash": "3647dc307f207768",
              "camera_direction_hash": "fa48c53d112ab7f2",
              "owned_usage_hash": "0dce98ebed01fd41",
              "validator": "scene_detail_owned_objects.v1",
              "violations": [
                {
                  "owned_object": "tunnel",
                  "violating_phrase": "weathered tunnel floor",
                  "reason": "The prompt references the tunnel floor purely as a spatial anchor/margin around the central phone without requesting redraw, matching the producer declaration.",
                  "verdict": "anchor_reference"
                }
              ]
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": -1,
        "dependency_reason": "",
        "_shot_index": 33,
        "visible_entities": [
          "P01"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 4,
            "shot_index": 33
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "insert",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "Directly above the smartphone and aligned with its screen plane, the camera holds completely still as the display advances to 21:14:08. The screen sits centrally with its bezel and a small surrounding margin retained, presenting both the changed time and the 17 percent battery reading without distorting the phone's physical scale.",
            "lighting_mood": "The active display supplies localized screen illumination within an otherwise restrained low-key frame.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 0명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "primary_subject_only_fully_visible": "the named primary subject (hand / face / object) is the only fully visible entity in the frame",
                  "other_entities_appearance_options": [
                    "partial visibility — shoulder slice / arm slice / sleeve / fingertips entering the frame edge",
                    "soft background — out-of-focus silhouette via shallow depth of field",
                    "absent — not rendered at all even if listed in visible_entities"
                  ],
                  "forbidden_combinations": [
                    "same character's full body and body-part close-up in the same frame (view mixing)",
                    "two characters both rendered as sharp close-up faces in the same frame"
                  ],
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "required_background_position",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "smartphone display showing 21:14:08 and 17 percent",
                  "screen_zone": "middle_center",
                  "depth_plane": "foreground",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "P01"
            ],
            "allowed_outlook_pairs": [],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": true,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "background_ref_attached",
            "bg_id": "L02B01",
            "reference_usage": "exact_background",
            "owned_objects": [
              "blast door",
              "light",
              "mine portal",
              "root",
              "sensor",
              "threshold",
              "tunnel",
              "wall"
            ],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "owned_objects lists all environment objects already drawn in the background reference — do not create new ones",
              "if focusing on an owned object, declare its reference-kind via the reference_phrase_kinds sidecar (producer contract)"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "tunnel_low_key_lighting",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The surrounding environment is a dark, ruined underground tunnel. Restrained low-key illumination dominates the space, preserving deep shadows, mineral blacks, and weathered neutral tones. This consistent lighting condition creates a somber, uncertain, and oppressive mood throughout the area, enveloping all subjects and objects.",
                "element_scope": "full",
                "applies_to_shots": [
                  32,
                  33,
                  36
                ],
                "source_facts": [
                  "preserving the dark, uncertain tunnel mood without naming an unstated source",
                  "an otherwise restrained low-key frame",
                  "Low-to-mid-key ambient light appropriate to the tunnel preserves restrained mineral blacks and weathered neutral tones"
                ],
                "visual_inferences": [
                  "The lack of a prominent light source means deep shadows naturally wrap around the periphery of all elements, seamlessly tying the separate shots into a single dark location."
                ],
                "creative_decisions": [
                  "Maintain a unified low-key exposure and shadow depth across different subjects (the face, the phone screen, the roots) to firmly anchor them in the same collapsed tunnel environment."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 4,
                "shot_index": 32,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "background",
                "id": "L02B01",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "e92d9330606d88ba"
      },
      {
        "scene_index": 4,
        "heading": "S4. INT. 같은 갱도 — 알 수 없는 시간 (01:50–02:35)",
        "beat_title": "변형된 갱도 인지",
        "representative_moment": "거대한 나무뿌리가 콘크리트와 철문을 찢고 나온 갱도의 기이한 풍경.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "Knee-height observer POV, wide upward angle, low-to-mid-key ambient lighting",
            "t2i_prompt": "Photorealistic cinematic still. [L02: A dark, ruined underground tunnel wrapped in deep shadows, lit by restrained low-to-mid-key ambient light preserving mineral blacks and weathered neutral tones.] Seen from knee height and angled upward across the damaged section filling the wide frame. The perspective frames the massive existing root as it passes diagonally across the center, situated between the broken opening of the split concrete wall on one side and the existing torn blast door on the opposing side. Deep shadows wrap around the periphery of all elements, seamlessly anchoring them in the somber environment.",
            "outfit_assignments": [],
            "source_facts": [
              "사람 허리만 한 나무뿌리가 콘크리트와 철문을 찢어 놓았다.",
              "Shot description: 갈라진 콘크리트와 찢겨진 철문 사이를 뱀처럼 휘감은 거대한 나무뿌리."
            ],
            "visual_inferences": [
              "fixed_element tunnel_low_key_lighting: Restrained low-key illumination dominates the space, preserving deep shadows, mineral blacks, and weathered neutral tones.",
              "The lateral tracking movement described in the camera direction is frozen as a wide composition holding the diagonal curve of the root between the concrete and door."
            ],
            "creative_decisions": [
              "Positioned the camera at knee height pointing upward to emphasize the massive scale and destructive intrusion of the root into the man-made corridor.",
              "Utilized deep shadows around the frame edges to focus attention on the central damaged structures."
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [],
            "reference_phrase_kinds": [
              "background"
            ],
            "owned_object_usage": [
              {
                "owned_token": "blast door",
                "usage_kind": "anchor",
                "source_phrase": "existing torn blast door on the opposing side"
              },
              {
                "owned_token": "light",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "mine portal",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "root",
                "usage_kind": "anchor",
                "source_phrase": "frames the massive existing root as it passes diagonally across the center"
              },
              {
                "owned_token": "sensor",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "threshold",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tunnel",
                "usage_kind": "anchor",
                "source_phrase": "[L02: A dark, ruined underground tunnel wrapped in deep shadows"
              },
              {
                "owned_token": "wall",
                "usage_kind": "anchor",
                "source_phrase": "broken opening of the split concrete wall on one side"
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "61e244a77e565008",
              "owned_hash": "3647dc307f207768",
              "camera_direction_hash": "de330c6c26672617",
              "owned_usage_hash": "653d546ebb007a8d",
              "validator": "scene_detail_owned_objects.v1",
              "violations": [
                {
                  "owned_object": "blast door",
                  "violating_phrase": "existing torn blast door on the opposing side",
                  "reason": "The prompt references the existing blast door purely as a framing and spatial landmark without instructing any new creation or modification, consistent with the producer's anchor declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "root",
                  "violating_phrase": "frames the massive existing root as it passes diagonally across the center",
                  "reason": "The prompt treats the massive root as an already existing central composition anchor across the frame, matching the producer's anchor declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "tunnel",
                  "violating_phrase": "A dark, ruined underground tunnel wrapped in deep shadows",
                  "reason": "The prompt mentions the tunnel strictly as the existing environmental setting context rather than commanding a new background structure, matching the producer's anchor declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "wall",
                  "violating_phrase": "broken opening of the split concrete wall on one side",
                  "reason": "The split concrete wall is described solely as a spatial boundary and compositional anchor without requesting new generation, matching the producer's anchor declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "light",
                  "violating_phrase": "lit by restrained low-to-mid-key ambient light",
                  "reason": "The phrase mentions ambient light purely as a global illumination descriptor without adding or redesigning any light fixture, functioning as an anchor reference despite producer's absent declaration.",
                  "verdict": "anchor_reference"
                }
              ]
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": 4,
        "dependency_reason": "Shot 33의 공간적 연속성 및 동일한 갱도 환경 유지 (exact_background)",
        "_shot_index": 36,
        "visible_entities": [
          "L02"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 4,
            "shot_index": 36
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "wide",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "From knee height just beyond Tony's former rear-side path, the camera continues a restrained lateral track, angled upward across the damaged section. The enormous root curves diagonally through the middle of frame between split concrete on one side and the torn iron door on the other, with the movement easing into a brief hold that makes its winding route legible.",
            "lighting_mood": "Low-to-mid-key ambient light appropriate to the tunnel preserves restrained mineral blacks and weathered neutral tones.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 0명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=wide — close 규칙은 이 샷에 해당하지 않는다"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": null,
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "L02"
            ],
            "allowed_outlook_pairs": [],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "background_ref_attached",
            "bg_id": "L02B01",
            "reference_usage": "exact_background",
            "owned_objects": [
              "blast door",
              "light",
              "mine portal",
              "root",
              "sensor",
              "threshold",
              "tunnel",
              "wall"
            ],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "owned_objects lists all environment objects already drawn in the background reference — do not create new ones",
              "if focusing on an owned object, declare its reference-kind via the reference_phrase_kinds sidecar (producer contract)"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "tunnel_low_key_lighting",
                "element_type": "environment_state",
                "character_name": "",
                "description": "The surrounding environment is a dark, ruined underground tunnel. Restrained low-key illumination dominates the space, preserving deep shadows, mineral blacks, and weathered neutral tones. This consistent lighting condition creates a somber, uncertain, and oppressive mood throughout the area, enveloping all subjects and objects.",
                "element_scope": "full",
                "applies_to_shots": [
                  32,
                  33,
                  36
                ],
                "source_facts": [
                  "preserving the dark, uncertain tunnel mood without naming an unstated source",
                  "an otherwise restrained low-key frame",
                  "Low-to-mid-key ambient light appropriate to the tunnel preserves restrained mineral blacks and weathered neutral tones"
                ],
                "visual_inferences": [
                  "The lack of a prominent light source means deep shadows naturally wrap around the periphery of all elements, seamlessly tying the separate shots into a single dark location."
                ],
                "creative_decisions": [
                  "Maintain a unified low-key exposure and shadow depth across different subjects (the face, the phone screen, the roots) to firmly anchor them in the same collapsed tunnel environment."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 4,
                "shot_index": 33,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "background",
                "id": "L02B01",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "eadac0e111409b79"
      },
      {
        "scene_index": 5,
        "heading": "S5. EXT. 갱도 출구 / 펜실베이니아 숲 — 낮",
        "beat_title": "사라진 인프라와 숲에 묻힌 세계",
        "representative_moment": "울창한 수관 위로 위태롭게 기울어진 콘크리트 고층건물의 뼈대가 광활한 원시 숲 한가운데 솟아 있다.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "elevated wide long-shot distance, grounded contrast, strong clear daylight",
            "t2i_prompt": "Photorealistic cinematic still. [L03B01: Seen from an elevated long-shot distance oblique to the valley axis, a dense forest canopy dominates the middle-center background, providing a vast horizontal scale reference. A weathered leaning high-rise skeleton protrudes precariously in the upper-center background, its skeletal concrete frame tilting diagonally to interrupt the horizontal landscape. The expansive primeval forest stretches unbroken across the terrain, bisected by a river flowing through the valley. Strong, clear daylight renders the wild environment with grounded contrast, highlighting the restrained greens of the foliage and the weathered neutral tones of the ruined structure.]",
            "outfit_assignments": [],
            "source_facts": [
              "끝없는 숲. 계곡을 가로지르는 강.",
              "멀리 나무 위로 비스듬히 솟은 콘크리트 고층건물의 골조.",
              "Shot 41 description: 울창한 수관 위로 위태롭게 기울어진 채 솟아 있는 콘크리트 고층건물의 뼈대."
            ],
            "visual_inferences": [
              "fixed_element leaning_highrise_ruin: skeletal structure of a concrete high-rise, completely devoid of glass or modern cladding, leaning precariously at a diagonal angle.",
              "fixed_element primeval_forest_landscape: immense overgrown forest lacking modern infrastructure, stretching across the terrain under strong daylight."
            ],
            "creative_decisions": [
              "Elevated wide long-shot framing to establish the vast scale of the unbroken forest against the diagonally leaning skeletal building.",
              "Applied strong, grounded daylight with a natural palette of restrained greens and weathered neutrals to emphasize the post-apocalyptic, overgrown environment without human presence."
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001",
              "fsc_002"
            ],
            "reference_phrase_kinds": [
              "background"
            ],
            "owned_object_usage": [
              {
                "owned_token": "mine portal",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rail",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "river",
                "usage_kind": "anchor",
                "source_phrase": "bisected by a river flowing through the valley"
              },
              {
                "owned_token": "support beam",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tower",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "vine",
                "usage_kind": "absent",
                "source_phrase": ""
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "0d1486f0e54e0387",
              "owned_hash": "3f3d4ee4c229af1f",
              "camera_direction_hash": "a388f23514d087c7",
              "owned_usage_hash": "cdada21866e89ae6",
              "validator": "scene_detail_owned_objects.v1",
              "violations": [
                {
                  "owned_object": "river",
                  "violating_phrase": "bisected by a river flowing through the valley",
                  "reason": "The river is mentioned purely as an existing environmental anchor describing the terrain, matching the producer's anchor declaration.",
                  "verdict": "anchor_reference"
                }
              ]
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": -1,
        "dependency_reason": "",
        "_shot_index": 41,
        "visible_entities": [
          "L03"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 5,
            "shot_index": 41
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "wide",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "At elevated long-shot distance and oblique to the valley axis, the pan eases across the treetops and ends with the leaning high-rise skeleton isolated in the upper portion of frame. The canopy remains broad beneath it as a scale reference, while the building's diagonal posture interrupts the otherwise horizontal landscape.",
            "lighting_mood": "Strong daylight renders the forest and distant concrete structure with grounded contrast and a restrained natural palette.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 0명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=wide — close 규칙은 이 샷에 해당하지 않는다"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "required_background_position",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "forest canopy",
                  "screen_zone": "middle_center",
                  "depth_plane": "background",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                },
                {
                  "constraint_id": "fsc_002",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "leaning high-rise skeleton",
                  "screen_zone": "upper_center",
                  "depth_plane": "background",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "L03"
            ],
            "allowed_outlook_pairs": [],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "background_ref_attached",
            "bg_id": "L03B01",
            "reference_usage": "exact_background",
            "owned_objects": [
              "mine portal",
              "rail",
              "river",
              "support beam",
              "tower",
              "vine"
            ],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "owned_objects lists all environment objects already drawn in the background reference — do not create new ones",
              "if focusing on an owned object, declare its reference-kind via the reference_phrase_kinds sidecar (producer contract)"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "leaning_highrise_ruin",
                "element_type": "environment_state",
                "character_name": "",
                "description": "A weathered, skeletal structure of a concrete high-rise building protruding above a dense forest canopy. It leans precariously at a diagonal angle, contrasting sharply with the natural, horizontal landscape of the surrounding primeval woods and river valley under strong, clear daylight.",
                "element_scope": "full",
                "applies_to_shots": [
                  41,
                  43
                ],
                "source_facts": [
                  "멀리 나무 위로 비스듬히 솟은 콘크리트 고층건물의 골조",
                  "leaning high-rise skeleton isolated in the upper portion of frame",
                  "distant leaning structure spread across the greater part of the composition"
                ],
                "visual_inferences": [
                  "The concrete structure is heavily weathered and completely devoid of glass or modern cladding, indicating centuries of abandonment."
                ],
                "creative_decisions": [
                  "The diagonal tilt of the ruined building serves as a stark visual anchor, emphasizing the collapse of the old civilization against the overpowering nature."
                ],
                "confidence": "high"
              },
              {
                "element_id": "primeval_forest_landscape",
                "element_type": "environment_state",
                "character_name": "",
                "description": "An expansive, immensely overgrown primeval forest stretching across the terrain and bisected by a river valley. The landscape completely lacks any modern infrastructure like highways or industrial complexes, bathed in strong daylight that highlights restrained greens and weathered neutral tones.",
                "element_scope": "full",
                "applies_to_shots": [
                  41,
                  43
                ],
                "source_facts": [
                  "끝없는 숲. 계곡을 가로지르는 강.",
                  "고속도로와 산업 단지는 사라졌다.",
                  "울창한 수관",
                  "Strong daylight renders the forest",
                  "retaining restrained greens and weathered neutrals"
                ],
                "visual_inferences": [
                  "The forest forms a thick, unbroken green carpet that has swallowed all previous human civilization."
                ],
                "creative_decisions": [
                  "The vastness of the unbroken forest is used to visually isolate the protagonist and establish the absolute erasure of the world he once knew."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "background",
                "id": "L03B01",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "9452399528d66595"
      },
      {
        "scene_index": 5,
        "heading": "EXT. 갱도 출구 / 펜실베이니아 숲 — 낮",
        "beat_title": "문명의 붕괴 확인",
        "representative_moment": "토니(C01)가 통신이 끊긴 채 폐허가 된 무한한 숲과 기울어진 고층건물을 멍하니 바라본다.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "over-the-shoulder wide framing, strong clear daylight",
            "t2i_prompt": "Photorealistic cinematic still. C01O01 in a dark rescue work uniform, a Caucasian man in his 30s, stands in the middle-left midground with his back to the camera, shoulders slack and head forward, looking toward the expansive forest and valley in the middle-center background. Captured from slightly above shoulder height behind him, his small, motionless figure occupies the near field, back and shoulder volume preserved. Spread across the greater part of the composition is an expansive, immensely overgrown primeval forest bisected by a river valley, completely lacking any modern infrastructure. In the distance, a weathered, skeletal structure of a concrete high-rise building protrudes above the dense forest canopy, leaning precariously at a diagonal angle. Strong, clear daylight keeps the figure clearly separated against the vast landscape while highlighting restrained greens and weathered neutral tones.",
            "outfit_assignments": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "source_facts": [
              "토니가 광대한 숲을 바라본다.",
              "끝없는 숲. 계곡을 가로지르는 강. 멀리 나무 위로 비스듬히 솟은 콘크리트 고층건물의 골조."
            ],
            "visual_inferences": [
              "fixed_element leaning_highrise_ruin: weathered, skeletal structure of a concrete high-rise building leaning precariously at a diagonal angle",
              "fixed_element primeval_forest_landscape: expansive immensely overgrown primeval forest completely lacking modern infrastructure"
            ],
            "creative_decisions": [
              "Wide framing positioning the figure small in the near field to emphasize the scale of the collapsed civilization against the overpowering nature",
              "Over-the-shoulder perspective from slightly above to effectively anchor the character's viewpoint toward the vast transformed landscape",
              "Strong daylight with restrained greens to support the tense post-apocalyptic survival tone"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001",
              "fsc_002"
            ],
            "reference_phrase_kinds": [
              "character",
              "background"
            ],
            "owned_object_usage": [
              {
                "owned_token": "mine portal",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rail",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "river",
                "usage_kind": "anchor",
                "source_phrase": "bisected by a river valley"
              },
              {
                "owned_token": "support beam",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tower",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "vine",
                "usage_kind": "absent",
                "source_phrase": ""
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "2996bdb781a1b5a1",
              "owned_hash": "3f3d4ee4c229af1f",
              "camera_direction_hash": "5dbe2d9d22abdee1",
              "owned_usage_hash": "21f72cc275cdfd28",
              "validator": "scene_detail_owned_objects.v1",
              "violations": [
                {
                  "owned_object": "river",
                  "violating_phrase": "bisected by a river valley",
                  "reason": "The prompt mentions 'river valley' purely as a compositional landscape reference without directing any redraw or visual alteration, matching the producer's anchor declaration.",
                  "verdict": "anchor_reference"
                }
              ]
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": -1,
        "dependency_reason": "",
        "_shot_index": 43,
        "visible_entities": [
          "C01",
          "L03"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 5,
            "shot_index": 43
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "wide",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "The descending crane settles a few meters behind Tony at slightly above shoulder height, offset so his back occupies one side rather than the center. His small, motionless figure remains in the near field while the forest, river valley, and distant leaning structure spread across the greater part of the composition; his attention is fixed on that transformed landscape.",
            "lighting_mood": "Strong daylight keeps Tony clearly separated against the expansive forest while retaining restrained greens and weathered neutrals.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 1명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=wide — close 규칙은 이 샷에 해당하지 않는다"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "shared_space_relation",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "expansive forest and valley",
                  "screen_zone": "middle_center",
                  "depth_plane": "background",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                },
                {
                  "constraint_id": "fsc_002",
                  "target_kind": "character",
                  "target_id": "C01",
                  "label": "토니(앤서니 로저스)",
                  "screen_zone": "middle_left",
                  "depth_plane": "midground",
                  "gesture_action": "looks_toward",
                  "gesture_target_label": "expansive forest and valley"
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "C01",
              "L03"
            ],
            "allowed_outlook_pairs": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "background_ref_attached",
            "bg_id": "L03B01",
            "reference_usage": "exact_background",
            "owned_objects": [
              "mine portal",
              "rail",
              "river",
              "support beam",
              "tower",
              "vine"
            ],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "owned_objects lists all environment objects already drawn in the background reference — do not create new ones",
              "if focusing on an owned object, declare its reference-kind via the reference_phrase_kinds sidecar (producer contract)"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "leaning_highrise_ruin",
                "element_type": "environment_state",
                "character_name": "",
                "description": "A weathered, skeletal structure of a concrete high-rise building protruding above a dense forest canopy. It leans precariously at a diagonal angle, contrasting sharply with the natural, horizontal landscape of the surrounding primeval woods and river valley under strong, clear daylight.",
                "element_scope": "full",
                "applies_to_shots": [
                  41,
                  43
                ],
                "source_facts": [
                  "멀리 나무 위로 비스듬히 솟은 콘크리트 고층건물의 골조",
                  "leaning high-rise skeleton isolated in the upper portion of frame",
                  "distant leaning structure spread across the greater part of the composition"
                ],
                "visual_inferences": [
                  "The concrete structure is heavily weathered and completely devoid of glass or modern cladding, indicating centuries of abandonment."
                ],
                "creative_decisions": [
                  "The diagonal tilt of the ruined building serves as a stark visual anchor, emphasizing the collapse of the old civilization against the overpowering nature."
                ],
                "confidence": "high"
              },
              {
                "element_id": "primeval_forest_landscape",
                "element_type": "environment_state",
                "character_name": "",
                "description": "An expansive, immensely overgrown primeval forest stretching across the terrain and bisected by a river valley. The landscape completely lacks any modern infrastructure like highways or industrial complexes, bathed in strong daylight that highlights restrained greens and weathered neutral tones.",
                "element_scope": "full",
                "applies_to_shots": [
                  41,
                  43
                ],
                "source_facts": [
                  "끝없는 숲. 계곡을 가로지르는 강.",
                  "고속도로와 산업 단지는 사라졌다.",
                  "울창한 수관",
                  "Strong daylight renders the forest",
                  "retaining restrained greens and weathered neutrals"
                ],
                "visual_inferences": [
                  "The forest forms a thick, unbroken green carpet that has swallowed all previous human civilization."
                ],
                "creative_decisions": [
                  "The vastness of the unbroken forest is used to visually isolate the protagonist and establish the absolute erasure of the world he once knew."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 5,
                "shot_index": 41,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "character_outlook",
                "id": "C01O01",
                "policy": "required"
              },
              {
                "kind": "background",
                "id": "L03B01",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "3c219e9a8ab72b2a"
      },
      {
        "scene_index": 6,
        "heading": "EXT. 갱도 출구 / 펜실베이니아 숲 — 낮",
        "beat_title": "거대한 원통형 폐허의 발견",
        "representative_moment": "토니가 원통형으로 둥글게 파여 나간 기이한 건물 잔해를 올려다보며 멈춰 선다.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "wide angle from waist height, low-to-mid-key cinematic contrast",
            "t2i_prompt": "Photorealistic cinematic still. C01O01, a Caucasian man in his 30s in a dark rescue work uniform, is positioned in the middle-left midground, halted mid-step with his chin lifted at a three-quarter angle from the left, looking toward the immense cylindrical void in the middle-right background. Captured from a waist-height camera perspective behind and to one side of the figure, the composition looks past his profile. [L03B02: An exterior forest environment where an immense smooth cylindrical void is carved away from the building remains and concrete foundation. Trees, fern, and moss cover the wall and rubble under natural daylight, presenting restrained forest greens, weathered concrete neutrals, and a low-to-mid-key cinematic contrast.]",
            "outfit_assignments": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "source_facts": [
              "Shot 49 description: Tony looking up at a massive cylindrical section carved out of a ruined building.",
              "DP direction: camera POV observer from waist height behind and to one side of Tony, looking past his three-quarter profile.",
              "Daylight presents the discovery in restrained forest greens and concrete neutrals."
            ],
            "visual_inferences": [
              "Waist-height camera perspective firmly anchors the immense scale of the background void against the human subject.",
              "fixed_element forest_daylight_environment: maintaining natural lighting throughout the scene."
            ],
            "creative_decisions": [
              "Placed C01O01 in the middle-left midground and the void in the middle-right background to rigorously fulfill the frame spatial contract.",
              "Anchored background owned objects like fern, moss, wall, and rubble to perfectly integrate the L03B02 environment."
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001",
              "fsc_002"
            ],
            "reference_phrase_kinds": [
              "character",
              "background"
            ],
            "owned_object_usage": [
              {
                "owned_token": "beam",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "building",
                "usage_kind": "anchor",
                "source_phrase": "building remains"
              },
              {
                "owned_token": "doorway",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "fern",
                "usage_kind": "anchor",
                "source_phrase": "Trees, fern, and moss"
              },
              {
                "owned_token": "foundation",
                "usage_kind": "anchor",
                "source_phrase": "concrete foundation"
              },
              {
                "owned_token": "moss",
                "usage_kind": "anchor",
                "source_phrase": "Trees, fern, and moss"
              },
              {
                "owned_token": "rebar",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rubble",
                "usage_kind": "anchor",
                "source_phrase": "wall and rubble"
              },
              {
                "owned_token": "tree",
                "usage_kind": "anchor",
                "source_phrase": "Trees, fern, and moss"
              },
              {
                "owned_token": "vine",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "void",
                "usage_kind": "anchor",
                "source_phrase": "immense cylindrical void"
              },
              {
                "owned_token": "wall",
                "usage_kind": "anchor",
                "source_phrase": "wall and rubble"
              },
              {
                "owned_token": "window",
                "usage_kind": "absent",
                "source_phrase": ""
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "b57c8e965b52a598",
              "owned_hash": "fcdee243b06f486e",
              "camera_direction_hash": "31e691174387feac",
              "owned_usage_hash": "632402f6b8a96b24",
              "validator": "scene_detail_owned_objects.v1",
              "violations": [
                {
                  "owned_object": "building",
                  "violating_phrase": "building remains",
                  "reason": "The building is referenced purely as background context and spatial anchor without instructions to redraw, matching the producer declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "fern",
                  "violating_phrase": "Trees, fern, and moss",
                  "reason": "Fern is mentioned as part of the existing environmental setting description, matching the producer anchor declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "foundation",
                  "violating_phrase": "concrete foundation",
                  "reason": "The foundation is cited as an existing scene anchor within the background reference, matching the producer declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "moss",
                  "violating_phrase": "Trees, fern, and moss",
                  "reason": "Moss is referenced as existing surface coverage in the background without new generation directives, matching the producer declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "rubble",
                  "violating_phrase": "wall and rubble",
                  "reason": "Rubble is cited purely as existing background geometry and texture, matching the producer declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "tree",
                  "violating_phrase": "Trees, fern, and moss",
                  "reason": "Trees are referenced as existing forest environment elements, matching the producer declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "void",
                  "violating_phrase": "immense cylindrical void",
                  "reason": "The void is mentioned as the visual focal point and gaze target in the existing background, matching the producer declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "wall",
                  "violating_phrase": "wall and rubble",
                  "reason": "The wall is mentioned as an existing background structure being covered by foliage, matching the producer declaration.",
                  "verdict": "anchor_reference"
                }
              ]
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": 5,
        "dependency_reason": "같은 숲 공간 및 시간적 연속성을 위해 이전 샷의 환경 상태 유지",
        "_shot_index": 49,
        "visible_entities": [
          "C01",
          "L03"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 6,
            "shot_index": 49
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "wide",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "From waist height several meters behind and to one side of Tony, the crane completes its upward reveal and settles, looking past his three-quarter profile toward the immense cylindrical section removed from the building remains. Tony is held near one lateral third, caught just after halting with his chin lifted, while the carved void occupies the opposing background without overwhelming his scale.",
            "lighting_mood": "Daylight presents the discovery in restrained forest greens, concrete neutrals, and low-to-mid-key cinematic contrast.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 1명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=wide — close 규칙은 이 샷에 해당하지 않는다"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "looks_to_anchor",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "cylindrical void",
                  "screen_zone": "middle_right",
                  "depth_plane": "background",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                },
                {
                  "constraint_id": "fsc_002",
                  "target_kind": "character",
                  "target_id": "C01",
                  "label": "토니(앤서니 로저스)",
                  "screen_zone": "middle_left",
                  "depth_plane": "midground",
                  "gesture_action": "looks_toward",
                  "gesture_target_label": "cylindrical void"
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "C01",
              "L03"
            ],
            "allowed_outlook_pairs": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "background_ref_attached",
            "bg_id": "L03B02",
            "reference_usage": "exact_background",
            "owned_objects": [
              "beam",
              "building",
              "doorway",
              "fern",
              "foundation",
              "moss",
              "rebar",
              "rubble",
              "tree",
              "vine",
              "void",
              "wall",
              "window"
            ],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "owned_objects lists all environment objects already drawn in the background reference — do not create new ones",
              "if focusing on an owned object, declare its reference-kind via the reference_phrase_kinds sidecar (producer contract)"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "forest_daylight_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "An exterior forest environment illuminated by natural daylight. The natural surroundings consistently feature a color palette dominated by restrained forest greens and weathered neutral tones.",
                "element_scope": "full",
                "applies_to_shots": [
                  49,
                  53
                ],
                "source_facts": [
                  "EXT. 갱도 출구 / 펜실베이니아 숲 — 낮",
                  "Shot 49: Daylight presents the discovery in restrained forest greens, concrete neutrals",
                  "Shot 53: Daylight is held to restrained forest greens and weathered neutrals"
                ],
                "visual_inferences": [
                  "The overall weather and ambient light in the forest remain steady, providing a consistent visual backdrop throughout the different days of the survival montage."
                ],
                "creative_decisions": [
                  "Maintaining a subdued, restrained color palette of greens and neutrals to ground the scene's realistic and tense survival atmosphere."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 5,
                "shot_index": 43,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "character_outlook",
                "id": "C01O01",
                "policy": "required"
              },
              {
                "kind": "background",
                "id": "L03B02",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "2f7e6d7f0b682419"
      },
      {
        "scene_index": 6,
        "heading": "EXT. 갱도 출구 / 펜실베이니아 숲 — 낮",
        "beat_title": "미지의 위협에 숨죽인 경계",
        "representative_moment": "토니(C01)가 부러진 나뭇가지 소리가 난 수풀 쪽을 향해 바닥에 바짝 엎드려 노려보고 있다.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "ground-level wide framing with strong diagonal composition and muted daylight",
            "t2i_prompt": "Photorealistic cinematic still. C01O01, a Caucasian man in his 30s in a black disaster rescue uniform, lies pressed prone in the lower-left midground, looking toward the brush where the sound originated in the upper-right background. Seen from a low observer at ground level to the side and slightly behind his feet, his boots and lower legs occupy the lower-left edge, his entire body cutting diagonally across the frame. Seen at a three-quarter angle from the right, his chest and hips are pressed low into the leaf litter, elbows braced, his head raised just enough to fix his intense stare off-screen. The scene is set in [L03: an exterior forest environment, a clear corridor of ground separating the figure from a distant shrub and fern, illuminated by natural daylight with restrained forest greens and weathered neutral tones, grounded contrast casting deep shadows under his body].",
            "outfit_assignments": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "source_facts": [
              "S6 action: '나뭇가지 부러지는 소리. 토니가 몸을 낮춘다.'",
              "Shot description: Tony prone on the ground, staring tensely toward a brush where a sound originated."
            ],
            "visual_inferences": [
              "fixed_element environment_state: natural daylight dominating restrained forest greens",
              "Tony maintains his prone posture tightly against the ground to avoid detection."
            ],
            "creative_decisions": [
              "Low ground-level framing emphasizes Tony's physical connection to the forest floor and his vulnerability.",
              "A strong diagonal composition from his boots to his head leads the viewer's eye toward the unseen threat.",
              "Muted daylight with grounded contrast enhances the sudden tension of the survival scene."
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001",
              "fsc_002"
            ],
            "reference_phrase_kinds": [
              "character",
              "background"
            ],
            "owned_object_usage": [
              {
                "owned_token": "drainage channel",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "fern",
                "usage_kind": "anchor",
                "source_phrase": "distant shrub and fern"
              },
              {
                "owned_token": "leaf litter",
                "usage_kind": "anchor",
                "source_phrase": "pressed low into the leaf litter"
              },
              {
                "owned_token": "mine entrance",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rail",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "rock",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "shrub",
                "usage_kind": "anchor",
                "source_phrase": "distant shrub and fern"
              },
              {
                "owned_token": "support beam",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "tree",
                "usage_kind": "absent",
                "source_phrase": ""
              },
              {
                "owned_token": "vine",
                "usage_kind": "absent",
                "source_phrase": ""
              }
            ],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "b82dab8d3bd40653",
              "owned_hash": "29751f863bbf5580",
              "camera_direction_hash": "eec328a1ac22d8bb",
              "owned_usage_hash": "b51eb4674848b42b",
              "validator": "scene_detail_owned_objects.v1",
              "violations": [
                {
                  "owned_object": "fern",
                  "violating_phrase": "distant shrub and fern",
                  "reason": "The prompt references the distant fern purely as a spatial background anchor, matching the producer's declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "leaf litter",
                  "violating_phrase": "pressed low into the leaf litter",
                  "reason": "The prompt references the leaf litter as the existing ground surface for the character's prone posture without requesting a redraw, matching the producer's declaration.",
                  "verdict": "anchor_reference"
                },
                {
                  "owned_object": "shrub",
                  "violating_phrase": "distant shrub and fern",
                  "reason": "The prompt mentions the distant shrub purely as a spatial composition landmark, matching the producer's declaration.",
                  "verdict": "anchor_reference"
                }
              ]
            }
          }
        ],
        "scene_type": "montage",
        "dependent_scene_index": 6,
        "dependency_reason": "S6_Shot49 exact_background를 참조하여 동일한 숲 환경의 연속성을 유지",
        "_shot_index": 53,
        "visible_entities": [
          "C01",
          "L03"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 6,
            "shot_index": 53
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "wide",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "At ground height several meters to Tony's side and slightly behind his feet, the lateral track stops with his entire prone body cutting diagonally from the lower foreground toward the brush. His chest and hips are pressed low, his head raised only enough to aim his stare toward the sound source, leaving a clear corridor of ground between him and the brush.",
            "lighting_mood": "Daylight is held to restrained forest greens and weathered neutrals, with grounded contrast supporting the sudden tension.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 1명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=wide — close 규칙은 이 샷에 해당하지 않는다"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "looks_to_anchor",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "brush where the sound originated",
                  "screen_zone": "upper_right",
                  "depth_plane": "background",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                },
                {
                  "constraint_id": "fsc_002",
                  "target_kind": "character",
                  "target_id": "C01",
                  "label": "토니(앤서니 로저스)",
                  "screen_zone": "lower_left",
                  "depth_plane": "midground",
                  "gesture_action": "looks_toward",
                  "gesture_target_label": "brush where the sound originated"
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "C01",
              "L03"
            ],
            "allowed_outlook_pairs": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "background_ref_attached",
            "bg_id": "L03B03",
            "reference_usage": "exact_background",
            "owned_objects": [
              "drainage channel",
              "fern",
              "leaf litter",
              "mine entrance",
              "rail",
              "rock",
              "shrub",
              "support beam",
              "tree",
              "vine"
            ],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "owned_objects lists all environment objects already drawn in the background reference — do not create new ones",
              "if focusing on an owned object, declare its reference-kind via the reference_phrase_kinds sidecar (producer contract)"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "forest_daylight_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "An exterior forest environment illuminated by natural daylight. The natural surroundings consistently feature a color palette dominated by restrained forest greens and weathered neutral tones.",
                "element_scope": "full",
                "applies_to_shots": [
                  49,
                  53
                ],
                "source_facts": [
                  "EXT. 갱도 출구 / 펜실베이니아 숲 — 낮",
                  "Shot 49: Daylight presents the discovery in restrained forest greens, concrete neutrals",
                  "Shot 53: Daylight is held to restrained forest greens and weathered neutrals"
                ],
                "visual_inferences": [
                  "The overall weather and ambient light in the forest remain steady, providing a consistent visual backdrop throughout the different days of the survival montage."
                ],
                "creative_decisions": [
                  "Maintaining a subdued, restrained color palette of greens and neutrals to ground the scene's realistic and tense survival atmosphere."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 6,
                "shot_index": 49,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "character_outlook",
                "id": "C01O01",
                "policy": "required"
              },
              {
                "kind": "background",
                "id": "L03B03",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "b8bcb6a3316a865d"
      },
      {
        "scene_index": 7,
        "heading": "S7. EXT. 숲속 능선 — 낮 (03:05–04:20)",
        "beat_title": "공중 도약과 기습 교전",
        "representative_moment": "C02가 점퍼 벨트를 이용해 거대한 나무 사이를 수평으로 도약하는 순간",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "wide lateral tracking shot slightly below flight level, crisp daylight separation",
            "t2i_prompt": "Photorealistic cinematic still. C02O02, a Caucasian woman in her late 20s in a rugged green flight suit, is airborne in the middle-center midground, moving toward the arrival tree at frame right in the middle-right background, with the departure tree visible in the middle-left background. Seen from slightly below her flight level in right profile at a wide full-body distance, she is suspended horizontally mid-action across the open gap. Both of her legs are extended forward, while the loose fabric of her suit trails strongly backward in the direction opposite to motion. A mechanical jumper belt is securely fastened directly under her armpits. Her eyes are looking off-screen. The lateral tracking composition focuses on her mid-air posture against the dense woodland foliage of the rugged forest ridge environment. The massive, ancient tree trunks frame her trajectory. Natural daylight filters through the canopy, providing crisp separation for the airborne figure within a restrained forest-green and weathered-neutral palette.",
            "outfit_assignments": [
              {
                "character_id": "C02",
                "outlook_id": "O02"
              }
            ],
            "source_facts": [
              "Shot description: 윌마 전신이 두 발을 뻗고 수평으로 도약해 허공에 뜬 mid-action 자세",
              "비행 반대 방향으로 옷자락이 강하게 휘날리는",
              "S7 action: 초록색 형체가 나무 위를 날아간다",
              "겨드랑이 아래의 점퍼 벨트가 희미하게 진동한다"
            ],
            "visual_inferences": [
              "fixed_element wilma_flight_attire: rugged green flight suit, mechanical jumper belt fastened directly under armpits",
              "fixed_element forest_ridge_environment: massive ancient tree trunks, natural daylight"
            ],
            "creative_decisions": [
              "wide lateral tracking framing to emphasize the horizontal gap and flight speed",
              "crisp daylight separation to distinguish the green suit from the dense forest background"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001",
              "fsc_002",
              "fsc_003"
            ],
            "reference_phrase_kinds": [
              "character"
            ],
            "owned_object_usage": [],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "289acf07919e3fc9",
              "owned_hash": "e3b0c44298fc1c14",
              "camera_direction_hash": "4d8f7032bc35e96f",
              "owned_usage_hash": "4f53cda18c2baa0c",
              "validator": "scene_detail_owned_objects.v1",
              "violations": []
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": -1,
        "dependency_reason": "No direct visual dependency established for the start of the scene.",
        "_shot_index": 1,
        "visible_entities": [
          "C02",
          "L04"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 7,
            "shot_index": 1
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "wide",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "High among the trees and slightly below Wilma's flight level, the camera tracks laterally at full-body distance as she crosses the open gap from frame left toward frame right. Her airborne body forms a horizontal line through the middle of frame, both legs extended toward the arrival direction while her clothing trails backward; the departure and arrival trees remain visible as spatial anchors.",
            "lighting_mood": "Daylight gives the airborne figure crisp separation within a restrained forest-green and weathered-neutral palette.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 1명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=wide — close 규칙은 이 샷에 해당하지 않는다"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "movement_direction",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "arrival tree",
                  "screen_zone": "middle_right",
                  "depth_plane": "background",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                },
                {
                  "constraint_id": "fsc_002",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "departure tree",
                  "screen_zone": "middle_left",
                  "depth_plane": "background",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                },
                {
                  "constraint_id": "fsc_003",
                  "target_kind": "character",
                  "target_id": "C02",
                  "label": "윌마 디어링",
                  "screen_zone": "middle_center",
                  "depth_plane": "midground",
                  "gesture_action": "moves_toward",
                  "gesture_target_label": "arrival tree at frame right"
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "C02",
              "L04"
            ],
            "allowed_outlook_pairs": [
              {
                "character_id": "C02",
                "outlook_id": "O02"
              }
            ],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "not_applicable",
            "bg_id": null,
            "reference_usage": "none",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "no chain_bg is bound to this shot — describe the location freely without referencing a background image"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "wilma_flight_attire",
                "element_type": "character_state",
                "character_name": "윌마 디어링",
                "description": "Wilma wears a predominantly green, rugged flight suit designed for high-speed aerial maneuvers. A mechanical jumper belt is securely fastened directly under her armpits, and her attire features loose fabric sections designed to catch the wind during movement.",
                "element_scope": "full",
                "applies_to_shots": [
                  1,
                  19
                ],
                "source_facts": [
                  "초록색 형체가 나무 위를 날아간다",
                  "겨드랑이 아래의 점퍼 벨트가 희미하게 진동한다",
                  "비행 반대 방향으로 옷자락이 강하게 휘날리는"
                ],
                "visual_inferences": [
                  "The description of a 'green figure' implies her functional flight suit is primarily a muted green to complement the forest surroundings."
                ],
                "creative_decisions": [
                  "The suit incorporates a weathered, utilitarian aesthetic, with the jumper belt serving as the prominent technological hardware on her upper torso."
                ],
                "confidence": "high"
              },
              {
                "element_id": "forest_ridge_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "A rugged forest ridge characterized by massive, ancient tree trunks and dense woodland foliage. The environment is illuminated by natural daylight, maintaining a restrained, weathered-neutral and forest-green color palette.",
                "element_scope": "full",
                "applies_to_shots": [
                  1,
                  14,
                  19
                ],
                "source_facts": [
                  "EXT. 숲속 능선 — 낮",
                  "거대 수목이 우거진 숲속 능선",
                  "restrained forest-green and weathered-neutral palette",
                  "forest remains faintly legible behind them",
                  "Daylight entering the hiding-place opening"
                ],
                "visual_inferences": [
                  "The tree trunks are exceptionally thick and towering, providing a spatial framework for aerial chases and clear separation for the foreground action."
                ],
                "creative_decisions": [
                  "Ambient daylight filters softly through the canopy, creating a slightly desaturated, natural backdrop that allows the dynamic foreground elements and explosions to stand out."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "character_outlook",
                "id": "C02O02",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "8b042bd48c2c0fc7"
      },
      {
        "scene_index": 7,
        "heading": "EXT. 숲속 능선 — 낮",
        "beat_title": "미지의 공중 폭발",
        "representative_moment": "A small flying drone shatters into pieces in midair, engulfed in an intense flame from a projectile impact.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "Close framing at observer head height, focused on the suspended midair fragmentation with fiery contrast against a soft background",
            "t2i_prompt": "Photorealistic cinematic still. A palm-sized flying drone, Midge, occupies the middle-center foreground, caught at the exact instant its mechanical body breaks apart in midair. Angled at three-quarter from the right, its front-facing camera points off-screen while its built-in inspection light is engulfed by an intense flame bursting from the core. Seen from an observer's head height and offset from the firing line, the central cluster of suspended fragments fills the close frame. The oblique projectile path cuts diagonally through the shattering debris, with peripheral fragments spreading toward the edges. Behind the high-contrast fiery impact, the daylight forest ridge is softly out of focus, its massive ancient tree trunks remaining faintly legible and oblique to the firing axis, preserving a restrained weathered-neutral and forest-green palette.",
            "outfit_assignments": [],
            "source_facts": [
              "S7 action: '폭발. 드론이 산산이 흩어진다.'",
              "shot description: '강렬한 불꽃과 함께 산산조각 나 허공에 흩뿌려진 파편들의 폭발 중간 순간의 미지(MIDGE).'",
              "camera_direction: 'At the armed pursuer's head height and offset from the firing axis, the static camera catches Midge at the instant its body breaks apart in midair.'"
            ],
            "visual_inferences": [
              "fixed_element environment_state: A rugged forest ridge characterized by massive, ancient tree trunks and dense woodland foliage.",
              "The drone's internal components and casing shatter radially along the projectile's diagonal trajectory."
            ],
            "creative_decisions": [
              "Close framing isolates the violent midair fragmentation, applying a shallow depth of field to keep the forest background soft and unobtrusive.",
              "The fiery explosion provides high-contrast illumination that highlights the sharp edges of the mechanical debris."
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001"
            ],
            "reference_phrase_kinds": [],
            "owned_object_usage": [],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "14264aa2afe495cf",
              "owned_hash": "e3b0c44298fc1c14",
              "camera_direction_hash": "f1f3efe3a40ae71b",
              "owned_usage_hash": "4f53cda18c2baa0c",
              "validator": "scene_detail_owned_objects.v1.close_skip",
              "violations": []
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": 7,
        "dependency_reason": "prior shot provides the established forest ridge environment which serves as the out-of-focus background for this close-up impact.",
        "_shot_index": 14,
        "visible_entities": [
          "C04"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 7,
            "shot_index": 14
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "close",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "At the armed pursuer's head height and offset from the firing axis, the static camera catches Midge at the instant its body breaks apart in midair. The central cluster of the drone and suspended fragments occupies most of the human-subject allowance without erasing the oblique projectile path, while peripheral fragments spread toward the frame edges and the forest remains faintly legible behind them.",
            "lighting_mood": "The impact's intense flame briefly drives the contrast against the daylight forest, while the surrounding palette remains restrained and rugged.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 1명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "primary_subject_only_fully_visible": "the named primary subject (hand / face / object) is the only fully visible entity in the frame",
                  "other_entities_appearance_options": [
                    "partial visibility — shoulder slice / arm slice / sleeve / fingertips entering the frame edge",
                    "soft background — out-of-focus silhouette via shallow depth of field",
                    "absent — not rendered at all even if listed in visible_entities"
                  ],
                  "forbidden_combinations": [
                    "same character's full body and body-part close-up in the same frame (view mixing)",
                    "two characters both rendered as sharp close-up faces in the same frame"
                  ],
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "wide_medium_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=close — wide/medium 규칙은 이 샷에 해당하지 않는다"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "primary_subject_isolation",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "character",
                  "target_id": "C04",
                  "label": "미지(MIDGE)",
                  "screen_zone": "middle_center",
                  "depth_plane": "foreground",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "C04"
            ],
            "allowed_outlook_pairs": [
              {
                "character_id": "C04",
                "outlook_id": "O00"
              }
            ],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [
              {
                "subject_id": "C04",
                "policy_type": "identity_reference",
                "policy": "generic_descriptor_allowed",
                "reason": "visible_shot_count=1 — selected-shot 저빈도 주변 subject"
              }
            ],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "skipped_close_framing",
            "bg_id": null,
            "reference_usage": "skipped_close_framing",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": true,
            "constraints": [
              "this shot uses close framing — the background reference is intentionally omitted at composition time",
              "describe the close subject and its immediate surroundings without relying on the absent background reference",
              "use [L##: ...] block for frame-edge surface description"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "forest_ridge_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "A rugged forest ridge characterized by massive, ancient tree trunks and dense woodland foliage. The environment is illuminated by natural daylight, maintaining a restrained, weathered-neutral and forest-green color palette.",
                "element_scope": "full",
                "applies_to_shots": [
                  1,
                  14,
                  19
                ],
                "source_facts": [
                  "EXT. 숲속 능선 — 낮",
                  "거대 수목이 우거진 숲속 능선",
                  "restrained forest-green and weathered-neutral palette",
                  "forest remains faintly legible behind them",
                  "Daylight entering the hiding-place opening"
                ],
                "visual_inferences": [
                  "The tree trunks are exceptionally thick and towering, providing a spatial framework for aerial chases and clear separation for the foreground action."
                ],
                "creative_decisions": [
                  "Ambient daylight filters softly through the canopy, creating a slightly desaturated, natural backdrop that allows the dynamic foreground elements and explosions to stand out."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 7,
                "shot_index": 1,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [],
            "forbidden_refs": [
              {
                "kind": "background",
                "reason": "close framing skips chain_bg reference"
              }
            ],
            "readiness_policy": "skipped_by_policy",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": true,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "b0f3631639201b16"
      },
      {
        "scene_index": 7,
        "heading": "EXT. 숲속 능선 — 낮",
        "beat_title": "은신처 발각과 대치",
        "representative_moment": "토니가 숨어 있는 은신처 안으로 윌마의 차가운 시선과 총구가 정확히 겨눠진 순간.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "Over-the-shoulder close framing from concealment, emphasizing the leveled weapon in the foreground against a softer midground.",
            "t2i_prompt": "Photorealistic cinematic still. C01O01 in a dark workwear suit, an American man in his 30s, stands pressed behind concealment in the middle-left foreground, looking toward C02O02. C02O02 in a predominantly green rugged flight suit, an American woman in her 20s, stands at a three-quarter angle from the left in the middle-center midground, looking toward C01O01 across the [: hiding-place opening, framed by rough edge surfaces] in the middle-center background. Seen from inside the concealment at C01O01's face height, the camera sits immediately beside his cheek and just off his shoulder, looking outward along the leveled barrel of P03. C01O01's cheek and tense eye hold the near left edge of the frame. The dark muzzle of P03 occupies a large foreground area, facing toward C01O01 and passing obliquely beside the camera. Behind the barrel, C02O02 is aligned with arms extended leveling the weapon, her suit displaying a mechanical jumper belt securely fastened directly under her armpits and loose fabric sections. Her eyes remain visibly focused in the softer midground behind the weapon. Daylight entering the opening creates controlled low-to-mid-key separation between the near muzzle, C01O01's face edge, and C02O02 beyond. Massive ancient tree trunks and dense woodland foliage maintain a restrained, weathered-neutral and forest-green color palette behind her.",
            "outfit_assignments": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              },
              {
                "character_id": "C02",
                "outlook_id": "O02"
              }
            ],
            "source_facts": [
              "S7 action: '윌마가 정확히 토니의 은신처로 총구를 돌린다.'",
              "shot description: 'Inside the hiding place at Tony's face height, the camera sits immediately beside his cheek and just off his shoulder, looking outward along the leveled gun barrel.'",
              "shot description: 'Tony's cheek and tense eye hold the near left edge, the muzzle occupies a large foreground area'",
              "shot description: 'Daylight entering the hiding-place opening creates controlled low-to-mid-key separation between the near muzzle, Tony's face edge, and Wilma beyond.'"
            ],
            "visual_inferences": [
              "fixed_element wilma_flight_attire: green rugged flight suit, mechanical jumper belt fastened under armpits, loose fabric sections",
              "fixed_element forest_ridge_environment: massive ancient tree trunks, weathered-neutral and forest-green palette"
            ],
            "creative_decisions": [
              "Applied a softer midground focus on Wilma to adhere to the close framing rule (preventing two sharp faces) while keeping her eye contact readable.",
              "Emphasized the muzzle of P03 dominating the foreground to convey the immediate threat passing beside the camera.",
              "Used L02 block to define the hiding-place opening as the shared spatial anchor bridging the foreground concealment and the midground confrontation."
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001",
              "fsc_002",
              "fsc_003"
            ],
            "reference_phrase_kinds": [
              "character",
              "prop"
            ],
            "owned_object_usage": [],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "d0eac882b567872d",
              "owned_hash": "e3b0c44298fc1c14",
              "camera_direction_hash": "ea7a458ae0e7a7e2",
              "owned_usage_hash": "4f53cda18c2baa0c",
              "validator": "scene_detail_owned_objects.v1.close_skip",
              "violations": []
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": -1,
        "dependency_reason": "",
        "_shot_index": 19,
        "visible_entities": [
          "C01",
          "C02",
          "P03"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 7,
            "shot_index": 19
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "close",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "Inside the hiding place at Tony's face height, the camera sits immediately beside his cheek and just off his shoulder, looking outward along the leveled gun barrel. Tony's cheek and tense eye hold the near left edge, the muzzle occupies a large but sub-two-fifths foreground area, and Wilma's cold eyes remain visible behind it at close distance as each fixes attention on the other.",
            "lighting_mood": "Daylight entering the hiding-place opening creates controlled low-to-mid-key separation between the near muzzle, Tony's face edge, and Wilma beyond.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)",
                "shared_anchor_required": "describe the shared surface or spatial anchor explicitly so both figures occupy the same physical space",
                "forbidden_phrasings": [
                  "fg/bg separation without naming the shared surface/space",
                  "isolated foreground hand reaching with another figure 'in the rear background' (no shared anchor named)",
                  "one figure 'in the foreground' and another 'in the back' without naming the room/surface they share"
                ],
                "recommended_keywords": [
                  "shared bench / shared table / shared surface",
                  "the same X they both occupy (X = bench, sofa, table, doorway, room)",
                  "across the X from each other",
                  "<prop> crosses the air between them (interaction prop visually links the two)",
                  "the <surface> sharing a single line of sight from the camera"
                ],
                "self_check_steps": [
                  "do the two characters share a single interaction (dialogue, handing, facing, co-watching)?",
                  "is one foreground and the other background-separated?",
                  "is the shared surface/space anchor explicitly named?",
                  "if not, add one of the recommended keywords"
                ],
                "rationale_summary": "fg/bg 분리만 묘사하면 모델이 두 인물을 다른 공간에 배치하거나 인터랙션이 끊긴 어색한 합성 — 공유 anchor 명시가 같은 물리적 공간 강제"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "primary_subject_only_fully_visible": "the named primary subject (hand / face / object) is the only fully visible entity in the frame",
                  "other_entities_appearance_options": [
                    "partial visibility — shoulder slice / arm slice / sleeve / fingertips entering the frame edge",
                    "soft background — out-of-focus silhouette via shallow depth of field",
                    "absent — not rendered at all even if listed in visible_entities"
                  ],
                  "forbidden_combinations": [
                    "same character's full body and body-part close-up in the same frame (view mixing)",
                    "two characters both rendered as sharp close-up faces in the same frame"
                  ],
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "wide_medium_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=close — wide/medium 규칙은 이 샷에 해당하지 않는다"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "shared_space_relation",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "hiding-place opening",
                  "screen_zone": "middle_center",
                  "depth_plane": "midground",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                },
                {
                  "constraint_id": "fsc_002",
                  "target_kind": "character",
                  "target_id": "C01",
                  "label": "토니(앤서니 로저스)",
                  "screen_zone": "middle_left",
                  "depth_plane": "foreground",
                  "gesture_action": "looks_toward",
                  "gesture_target_label": "윌마 디어링"
                },
                {
                  "constraint_id": "fsc_003",
                  "target_kind": "character",
                  "target_id": "C02",
                  "label": "윌마 디어링",
                  "screen_zone": "middle_center",
                  "depth_plane": "midground",
                  "gesture_action": "looks_toward",
                  "gesture_target_label": "토니(앤서니 로저스)"
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "two characters in fg/bg separation must share a named surface/space anchor — see render_strategy.spatial_consistency.fg_bg_shared_anchor_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "C01",
              "C02",
              "P03"
            ],
            "allowed_outlook_pairs": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              },
              {
                "character_id": "C02",
                "outlook_id": "O02"
              }
            ],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "skipped_close_framing",
            "bg_id": null,
            "reference_usage": "skipped_close_framing",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": true,
            "constraints": [
              "this shot uses close framing — the background reference is intentionally omitted at composition time",
              "describe the close subject and its immediate surroundings without relying on the absent background reference",
              "use [L##: ...] block for frame-edge surface description"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "wilma_flight_attire",
                "element_type": "character_state",
                "character_name": "윌마 디어링",
                "description": "Wilma wears a predominantly green, rugged flight suit designed for high-speed aerial maneuvers. A mechanical jumper belt is securely fastened directly under her armpits, and her attire features loose fabric sections designed to catch the wind during movement.",
                "element_scope": "full",
                "applies_to_shots": [
                  1,
                  19
                ],
                "source_facts": [
                  "초록색 형체가 나무 위를 날아간다",
                  "겨드랑이 아래의 점퍼 벨트가 희미하게 진동한다",
                  "비행 반대 방향으로 옷자락이 강하게 휘날리는"
                ],
                "visual_inferences": [
                  "The description of a 'green figure' implies her functional flight suit is primarily a muted green to complement the forest surroundings."
                ],
                "creative_decisions": [
                  "The suit incorporates a weathered, utilitarian aesthetic, with the jumper belt serving as the prominent technological hardware on her upper torso."
                ],
                "confidence": "high"
              },
              {
                "element_id": "forest_ridge_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "A rugged forest ridge characterized by massive, ancient tree trunks and dense woodland foliage. The environment is illuminated by natural daylight, maintaining a restrained, weathered-neutral and forest-green color palette.",
                "element_scope": "full",
                "applies_to_shots": [
                  1,
                  14,
                  19
                ],
                "source_facts": [
                  "EXT. 숲속 능선 — 낮",
                  "거대 수목이 우거진 숲속 능선",
                  "restrained forest-green and weathered-neutral palette",
                  "forest remains faintly legible behind them",
                  "Daylight entering the hiding-place opening"
                ],
                "visual_inferences": [
                  "The tree trunks are exceptionally thick and towering, providing a spatial framework for aerial chases and clear separation for the foreground action."
                ],
                "creative_decisions": [
                  "Ambient daylight filters softly through the canopy, creating a slightly desaturated, natural backdrop that allows the dynamic foreground elements and explosions to stand out."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 7,
                "shot_index": 1,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "character_outlook",
                "id": "C01O01",
                "policy": "required"
              },
              {
                "kind": "character_outlook",
                "id": "C02O02",
                "policy": "required"
              },
              {
                "kind": "prop",
                "id": "P03",
                "policy": "required"
              }
            ],
            "forbidden_refs": [
              {
                "kind": "background",
                "reason": "close framing skips chain_bg reference"
              }
            ],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [
            {
              "contract_id": "rc_001",
              "scope": {
                "scene_index": 7,
                "shot_indices": [
                  19
                ],
                "duration": "single_shot"
              },
              "targets": [
                {
                  "entity_id": "P03",
                  "role": "visual_target"
                }
              ],
              "requirements": [
                {
                  "dimension": "visual_identity",
                  "operation": "preserve",
                  "strength": "required",
                  "reference_policy": "use_entity_reference"
                }
              ]
            }
          ],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": true,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "74c221cae9a10c47"
      },
      {
        "scene_index": 8,
        "heading": "EXT. 숲속 능선 — 낮 (04:20–05:20)",
        "beat_title": "토니의 투항과 신원 확인",
        "representative_moment": "바닥에 구조용 칼을 내려놓은 토니(C01)가 양손을 들고 화면 밖의 상대를 응시한다.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "ground-level wide angle looking up, grounded contrast",
            "t2i_prompt": "Photorealistic cinematic still. Seen from ground level and tilted up along a metallic rescue knife that rests prominently on the earthy dirt in the near foreground. In the middle distance, C01O01 in dark rescue workwear, an American man in his mid-30s with a natural adult male face, stands at a three-quarter angle from the right. Both of his hands are raised at shoulder height in a gesture of surrender, his eyes fixed on someone off-screen. Behind him, the thick trunks of large forest trees rise, dominating the background. Restrained natural daylight supports a weathered-neutral and forest-green palette, casting soft shadows across the forest floor and maintaining grounded contrast around the surrender gesture.",
            "outfit_assignments": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "source_facts": [
              "EXT. 숲속 능선 — 낮 (04:20–05:20)",
              "토니가 구조용 칼을 바닥에 내려놓고 양손을 든 채 나온다.",
              "shot description: 토니가 바닥에 버려진 칼 곁에서 양손을 어깨 높이로 들어 올린 채 서 있다."
            ],
            "visual_inferences": [
              "fixed_element forest_ridge_environment: large trees, earthy dirt ground",
              "Tony looks at Wilma who is positioned off-screen",
              "low observer position near the knife creates an upward perspective line toward Tony"
            ],
            "creative_decisions": [
              "ground-level framing to anchor the foreground with the discarded weapon while placing Tony fully in the wide composition",
              "muted daylight to establish the tense, isolated atmosphere of the standoff"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [],
            "reference_phrase_kinds": [
              "character"
            ],
            "owned_object_usage": [],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "2a93b457cc7b47b7",
              "owned_hash": "e3b0c44298fc1c14",
              "camera_direction_hash": "402450cf5faeb873",
              "owned_usage_hash": "4f53cda18c2baa0c",
              "validator": "scene_detail_owned_objects.v1",
              "violations": []
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": 7,
        "dependency_reason": "앞선 7씬의 숲속 능선(forest_ridge_environment) 배경 공간 연속성 유지",
        "_shot_index": 1,
        "visible_entities": [
          "C01",
          "L04"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 8,
            "shot_index": 1
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "wide",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "The opening frame holds at ground level beside the discarded rescue knife, offset from Tony's frontal axis and looking upward along his raised arms. The knife remains prominent but below two-fifths of the foreground, while Tony's full body occupies the middle distance in a three-quarter view, both hands held at shoulder height and his attention fixed on Wilma outside the frame.",
            "lighting_mood": "Daylight supports a restrained forest-green and weathered-neutral palette with grounded contrast around the surrender gesture.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 1명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=wide — close 규칙은 이 샷에 해당하지 않는다"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": null,
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "C01",
              "L04"
            ],
            "allowed_outlook_pairs": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              }
            ],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "not_applicable",
            "bg_id": null,
            "reference_usage": "none",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "no chain_bg is bound to this shot — describe the location freely without referencing a background image"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "forest_ridge_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "An exterior forest ridge during the day, featuring large trees, earthy dirt ground, and a weathered-neutral color palette. The scene is illuminated by restrained, muted natural daylight that provides controlled visibility while keeping the surrounding context soft and grounded.",
                "element_scope": "full",
                "applies_to_shots": [
                  1,
                  6,
                  7
                ],
                "source_facts": [
                  "EXT. 숲속 능선 — 낮 (04:20–05:20)",
                  "거대 수목이 우거진 숲속 능선",
                  "Daylight supports a restrained forest-green and weathered-neutral palette",
                  "surrounding context falls soft",
                  "Restrained exterior daylight keeps the eye sockets firm but readable"
                ],
                "visual_inferences": [
                  "The environment has a dense canopy that diffuses the early morning sunlight, creating soft shadows.",
                  "The background retains deep green and earth tones even when out of focus."
                ],
                "creative_decisions": [
                  "A consistent low-to-mid-key contrast is maintained across all shots to emphasize the tense, isolated atmosphere of the confrontation."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 7,
                "shot_index": 1,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "character_outlook",
                "id": "C01O01",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "dc49ea4d0a090e53"
      },
      {
        "scene_index": 8,
        "heading": "EXT. 숲속 능선 — 낮 (04:20–05:20)",
        "beat_title": "The Dead Phone Awakes",
        "representative_moment": "An insert view of P01's cracked screen showing the impossible date and an unsent message, held between two implied figures.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "shallow depth of field over-the-shoulder insert, restrained daylight",
            "t2i_prompt": "Photorealistic cinematic still. Seen from immediately behind and slightly above an out-of-frame shoulder, at close hand height, a legible smartphone display of P01 occupies just under two-fifths of the middle-center foreground. The device, held by a weathered hand, has its content-bearing front face angled slightly upward toward the camera. The cracked glass screen sits squarely, keeping the digital date '2026.10.18', a battery indicator, and unfinished text sharply legible against the black casing. Beyond the sharp edges of the phone, the surrounding forest ridge environment falls soft into a shallow depth of field. Restrained natural daylight provides controlled visibility, blurring the deep green tree trunks and earthy dirt ground in the out-of-focus background.",
            "outfit_assignments": [],
            "source_facts": [
              "The cracked display sits squarely enough for “2026.10.18,” “3%,” and the unfinished message",
              "phone stays just under two-fifths of the composition and the surrounding context falls soft"
            ],
            "visual_inferences": [
              "fixed_element persistent_prop: cracked_smartphone",
              "fixed_element environment_state: forest_ridge_environment",
              "shallow depth of field obscures the forest background to focus entirely on the screen data"
            ],
            "creative_decisions": [
              "translating the over-shoulder camera instruction into an implied viewpoint to keep characters out of frame while maintaining spatial tension",
              "muted natural daylight keeps the screen readable without glare"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001"
            ],
            "reference_phrase_kinds": [
              "background"
            ],
            "owned_object_usage": [],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "2cf4768258c6f2ac",
              "owned_hash": "e3b0c44298fc1c14",
              "camera_direction_hash": "c35f75aba5bd46f1",
              "owned_usage_hash": "4f53cda18c2baa0c",
              "validator": "scene_detail_owned_objects.v1",
              "violations": []
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": 7,
        "dependency_reason": "Maintains the forest ridge environment established in the prior scene while focusing on the persistent smartphone prop.",
        "_shot_index": 6,
        "visible_entities": [
          "P01"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 8,
            "shot_index": 6
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "insert",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "From immediately behind and slightly above Wilma's shoulder position, with her body kept outside the frame, the dolly-in settles at close hand height over the phone held between her and Tony. The cracked display sits squarely enough for “2026.10.18,” “3%,” and the unfinished message “마라, 갱도가 무너졌어. 나는—” to remain sharply legible, while the phone stays just under two-fifths of the composition and the surrounding context falls soft.",
            "lighting_mood": "Daylight provides neutral, controlled visibility for the active display without introducing an additional specified light source.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 0명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "primary_subject_only_fully_visible": "the named primary subject (hand / face / object) is the only fully visible entity in the frame",
                  "other_entities_appearance_options": [
                    "partial visibility — shoulder slice / arm slice / sleeve / fingertips entering the frame edge",
                    "soft background — out-of-focus silhouette via shallow depth of field",
                    "absent — not rendered at all even if listed in visible_entities"
                  ],
                  "forbidden_combinations": [
                    "same character's full body and body-part close-up in the same frame (view mixing)",
                    "two characters both rendered as sharp close-up faces in the same frame"
                  ],
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "required_background_position",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "legible smartphone display",
                  "screen_zone": "middle_center",
                  "depth_plane": "foreground",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "P01"
            ],
            "allowed_outlook_pairs": [],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": true,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "not_applicable",
            "bg_id": null,
            "reference_usage": "none",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "no chain_bg is bound to this shot — describe the location freely without referencing a background image"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "forest_ridge_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "An exterior forest ridge during the day, featuring large trees, earthy dirt ground, and a weathered-neutral color palette. The scene is illuminated by restrained, muted natural daylight that provides controlled visibility while keeping the surrounding context soft and grounded.",
                "element_scope": "full",
                "applies_to_shots": [
                  1,
                  6,
                  7
                ],
                "source_facts": [
                  "EXT. 숲속 능선 — 낮 (04:20–05:20)",
                  "거대 수목이 우거진 숲속 능선",
                  "Daylight supports a restrained forest-green and weathered-neutral palette",
                  "surrounding context falls soft",
                  "Restrained exterior daylight keeps the eye sockets firm but readable"
                ],
                "visual_inferences": [
                  "The environment has a dense canopy that diffuses the early morning sunlight, creating soft shadows.",
                  "The background retains deep green and earth tones even when out of focus."
                ],
                "creative_decisions": [
                  "A consistent low-to-mid-key contrast is maintained across all shots to emphasize the tense, isolated atmosphere of the confrontation."
                ],
                "confidence": "high"
              },
              {
                "element_id": "cracked_smartphone",
                "element_type": "persistent_prop",
                "character_name": "",
                "description": "A handheld 2020s smartphone with a clearly cracked display screen. When visible from the front, the active screen shows the text '2026.10.18', a 3 percent battery indicator, and an unfinished message. From other angles, its physical dark edges are visible.",
                "element_scope": "close",
                "applies_to_shots": [
                  6,
                  7
                ],
                "source_facts": [
                  "토니가 금이 간 휴대폰을 꺼낸다.",
                  "화면에는 2026.10.18, 배터리 3퍼센트, 그리고 미전송 메시지가 보인다.",
                  "The cracked display sits squarely enough for 2026.10.18, 3%, and the unfinished message",
                  "a narrow, unreadable rear edge of the device remains soft at the lower foreground"
                ],
                "visual_inferences": [
                  "The device is a typical rectangular modern smartphone with visible physical damage like shattered glass.",
                  "The screen's backlight is bright enough to be readable against the muted daylight of the forest."
                ],
                "creative_decisions": [
                  "The phone is framed to bridge the focus between characters, appearing sharply in focus when displaying information and serving as a soft foreground element when focus shifts to facial expressions."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 7,
                "shot_index": 14,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [],
            "forbidden_refs": [],
            "readiness_policy": "not_applicable",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "657d6e8b1280b851"
      },
      {
        "scene_index": 8,
        "heading": "EXT. 숲속 능선 — 낮 (04:20–05:20)",
        "beat_title": "낯선 파형과 미래의 날짜",
        "representative_moment": "토니의 휴대폰 화면을 향해 시선이 내리꽂힌 윌마(C02)의 굳은 눈빛이 프레임을 가득 채운다.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "Tight close-up from slightly below eye line, three-quarter angle, low-to-mid-key natural daylight",
            "t2i_prompt": "Photorealistic cinematic still. C02O02 in a green combat uniform, an American woman in her late 20s, Wilma Deering, occupies the middle-center foreground. Observed from slightly below her eye line and off her frontal axis at a three-quarter angle from the left, her locked brow, fixed eyes, and the bridge of her nose fill the frame. Her chin is dipped, and her gaze is fixed sharply downward. A narrow, physical dark rear edge of a cracked smartphone sits soft and out of focus at the lower frame edge, its unseen display facing her. Restrained exterior daylight casts firm but readable shadows around her eye sockets, while the surrounding weathered-neutral forest environment dissolves into a soft blur.",
            "outfit_assignments": [
              {
                "character_id": "C02",
                "outlook_id": "O02"
              }
            ],
            "source_facts": [
              "S8 action: 윌마의 시선이 메시지에 잠깐 머문다.",
              "shot description: 윌마의 굳은 눈동자",
              "camera_direction: static camera sits slightly below Wilma's eye line, tightly framing her brow, fixed eyes, and the bridge of her nose in a three-quarter view."
            ],
            "visual_inferences": [
              "fixed_element cracked_smartphone: positioned below frame, rendering only its rear edge softly at the lower foreground",
              "forest_ridge_environment: omitted at composition time due to close framing, appearing only as an out-of-focus blur"
            ],
            "creative_decisions": [
              "Close framing emphasizes Wilma's intense focus by isolating her eyes and brow, making the phone a soft foreground anchor",
              "Restrained daylight maintains the tense mood while clearly defining her eye sockets"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001"
            ],
            "reference_phrase_kinds": [
              "character"
            ],
            "owned_object_usage": [],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "faf88e186ad4bd30",
              "owned_hash": "e3b0c44298fc1c14",
              "camera_direction_hash": "38ad95361367bb2f",
              "owned_usage_hash": "4f53cda18c2baa0c",
              "validator": "scene_detail_owned_objects.v1.close_skip",
              "violations": []
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": 7,
        "dependency_reason": "동일한 숲속 능선 로케이션의 이전 샷 배경 조명 및 시각적 분위기 참조",
        "_shot_index": 7,
        "visible_entities": [
          "C02"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 8,
            "shot_index": 7
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "close",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "From Tony's side of the exchange, the static camera sits slightly below Wilma's eye line and off her frontal axis, tightly framing her brow, fixed eyes, and the bridge of her nose in a three-quarter view. Her eyes drive sharply downward toward the phone below frame, while a narrow, unreadable rear edge of the device remains soft at the lower foreground; the emphasized change is the crossed gaze axis, with position and lighting held steady.",
            "lighting_mood": "Restrained exterior daylight keeps the eye sockets firm but readable, with low-to-mid-key contrast and muted natural color.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 1명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "primary_subject_only_fully_visible": "the named primary subject (hand / face / object) is the only fully visible entity in the frame",
                  "other_entities_appearance_options": [
                    "partial visibility — shoulder slice / arm slice / sleeve / fingertips entering the frame edge",
                    "soft background — out-of-focus silhouette via shallow depth of field",
                    "absent — not rendered at all even if listed in visible_entities"
                  ],
                  "forbidden_combinations": [
                    "same character's full body and body-part close-up in the same frame (view mixing)",
                    "two characters both rendered as sharp close-up faces in the same frame"
                  ],
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "wide_medium_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=close — wide/medium 규칙은 이 샷에 해당하지 않는다"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "primary_subject_isolation",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "character",
                  "target_id": "C02",
                  "label": "윌마 디어링",
                  "screen_zone": "middle_center",
                  "depth_plane": "foreground",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "C02"
            ],
            "allowed_outlook_pairs": [
              {
                "character_id": "C02",
                "outlook_id": "O02"
              }
            ],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "skipped_close_framing",
            "bg_id": null,
            "reference_usage": "skipped_close_framing",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": true,
            "constraints": [
              "this shot uses close framing — the background reference is intentionally omitted at composition time",
              "describe the close subject and its immediate surroundings without relying on the absent background reference",
              "use [L##: ...] block for frame-edge surface description"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "forest_ridge_environment",
                "element_type": "environment_state",
                "character_name": "",
                "description": "An exterior forest ridge during the day, featuring large trees, earthy dirt ground, and a weathered-neutral color palette. The scene is illuminated by restrained, muted natural daylight that provides controlled visibility while keeping the surrounding context soft and grounded.",
                "element_scope": "full",
                "applies_to_shots": [
                  1,
                  6,
                  7
                ],
                "source_facts": [
                  "EXT. 숲속 능선 — 낮 (04:20–05:20)",
                  "거대 수목이 우거진 숲속 능선",
                  "Daylight supports a restrained forest-green and weathered-neutral palette",
                  "surrounding context falls soft",
                  "Restrained exterior daylight keeps the eye sockets firm but readable"
                ],
                "visual_inferences": [
                  "The environment has a dense canopy that diffuses the early morning sunlight, creating soft shadows.",
                  "The background retains deep green and earth tones even when out of focus."
                ],
                "creative_decisions": [
                  "A consistent low-to-mid-key contrast is maintained across all shots to emphasize the tense, isolated atmosphere of the confrontation."
                ],
                "confidence": "high"
              },
              {
                "element_id": "cracked_smartphone",
                "element_type": "persistent_prop",
                "character_name": "",
                "description": "A handheld 2020s smartphone with a clearly cracked display screen. When visible from the front, the active screen shows the text '2026.10.18', a 3 percent battery indicator, and an unfinished message. From other angles, its physical dark edges are visible.",
                "element_scope": "close",
                "applies_to_shots": [
                  6,
                  7
                ],
                "source_facts": [
                  "토니가 금이 간 휴대폰을 꺼낸다.",
                  "화면에는 2026.10.18, 배터리 3퍼센트, 그리고 미전송 메시지가 보인다.",
                  "The cracked display sits squarely enough for 2026.10.18, 3%, and the unfinished message",
                  "a narrow, unreadable rear edge of the device remains soft at the lower foreground"
                ],
                "visual_inferences": [
                  "The device is a typical rectangular modern smartphone with visible physical damage like shattered glass.",
                  "The screen's backlight is bright enough to be readable against the muted daylight of the forest."
                ],
                "creative_decisions": [
                  "The phone is framed to bridge the focus between characters, appearing sharply in focus when displaying information and serving as a soft foreground element when focus shifts to facial expressions."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 7,
                "shot_index": 1,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "character_outlook",
                "id": "C02O02",
                "policy": "required"
              }
            ],
            "forbidden_refs": [
              {
                "kind": "background",
                "reason": "close framing skips chain_bg reference"
              }
            ],
            "readiness_policy": "skipped_by_policy",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": true,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "086bc27d322ea1c5"
      },
      {
        "scene_index": 9,
        "heading": "EXT. 숲속 능선 — 낮 (05:20–06:00)",
        "beat_title": "2419년 동기화 확인",
        "representative_moment": "스마트폰 화면에 2419.03.18 이라는 동기화된 날짜가 선명하게 떠 있다.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "macro insert shot, steep downward angle, crisp digital focus",
            "t2i_prompt": "Photorealistic cinematic still. Seen from a steep downward angle very close above the device, P01, a rectangular smartphone with a black casing and cracked glass touchscreen, occupies the central third of the frame. The illuminated front display faces directly upward, square and readable, showing a modern digital interface with the crisp synchronized numerals '2419.03.18'. The active screen is the sole sharp focal plane. Beneath and immediately around the edges of the device, a narrow area of the earthy ridge surface is visible, providing physical scale. The shot is lit by subdued early morning exterior daylight casting a restrained, neutral contrast over the environment. This soft ambient illumination balances against the glowing display, preserving muted earthy tones on the ground without introducing harsh directional shadows.",
            "outfit_assignments": [],
            "source_facts": [
              "Shot 3 description: 스마트폰 화면 클로즈업. 날짜가 빠르게 변한다. 2419.03.18",
              "shot staging camera direction: steep downward camera keeping the front display square enough to read while the phone remains contained within the central third of the frame"
            ],
            "visual_inferences": [
              "fixed_element early_morning_daylight: Subdued early morning exterior daylight casting a restrained, neutral contrast",
              "The smartphone rests over the earthy ground of the ridge, providing the narrow background texture required for scale"
            ],
            "creative_decisions": [
              "Replaced the exact measurement 'a few inches' with 'very close above' to accommodate T2I limitation with real-world dimensions",
              "Centered the device against a narrow margin of the natural terrain to ground the digital interface in the harsh reality of the forest ridge"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [],
            "reference_phrase_kinds": [],
            "owned_object_usage": [],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "81fdbdd85349d4e0",
              "owned_hash": "e3b0c44298fc1c14",
              "camera_direction_hash": "8f700910cb0dded5",
              "owned_usage_hash": "4f53cda18c2baa0c",
              "validator": "scene_detail_owned_objects.v1",
              "violations": []
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": -1,
        "dependency_reason": "",
        "_shot_index": 3,
        "visible_entities": [
          "P01"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 9,
            "shot_index": 3
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "insert",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "A steep downward camera a few inches above the smartphone completes its dolly-in as the changing numerals stop, keeping the front display square enough to read while the phone remains contained within the central third of the frame. The synchronized date “2419.03.18” is the sole sharp focal plane, with the ridge surface retained narrowly around the device for scale; only camera distance is accented as motion resolves into stillness.",
            "lighting_mood": "Exterior daylight is balanced against the active display so the synchronized date remains crisp without introducing an additional light source.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 0명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "primary_subject_only_fully_visible": "the named primary subject (hand / face / object) is the only fully visible entity in the frame",
                  "other_entities_appearance_options": [
                    "partial visibility — shoulder slice / arm slice / sleeve / fingertips entering the frame edge",
                    "soft background — out-of-focus silhouette via shallow depth of field",
                    "absent — not rendered at all even if listed in visible_entities"
                  ],
                  "forbidden_combinations": [
                    "same character's full body and body-part close-up in the same frame (view mixing)",
                    "two characters both rendered as sharp close-up faces in the same frame"
                  ],
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": null,
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "P01"
            ],
            "allowed_outlook_pairs": [],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": true,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "not_applicable",
            "bg_id": null,
            "reference_usage": "none",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "no chain_bg is bound to this shot — describe the location freely without referencing a background image"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "early_morning_daylight",
                "element_type": "environment_state",
                "character_name": "",
                "description": "Subdued early morning exterior daylight casting a restrained, neutral contrast over the environment. This natural lighting preserves the muted earthy tones of the surroundings without introducing harsh directional shadows.",
                "element_scope": "full",
                "applies_to_shots": [
                  3,
                  5,
                  13
                ],
                "source_facts": [
                  "Exterior daylight is balanced against the active display (Shot 3)",
                  "Daylight over the forest... preserving subdued greens and neutral contrast (Shot 5)",
                  "Exterior daylight retains rugged, restrained contrast (Shot 13)"
                ],
                "visual_inferences": [
                  "The early morning timeframe (05:20-06:00) provides soft, ambient illumination across both macro surface details and wide landscape views."
                ],
                "creative_decisions": [
                  "Maintain a consistent desaturated and pale atmospheric tone to unify the varying scales of the shots."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [],
            "forbidden_refs": [],
            "readiness_policy": "not_applicable",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "17cd886cc2a340ad"
      },
      {
        "scene_index": 9,
        "heading": "EXT. 숲속 능선 — 낮 (05:20–06:00)",
        "beat_title": "절대적 파괴의 강림",
        "representative_moment": "거대한 창백한 빛기둥이 숲 한가운데로 수직으로 꽂혀 내리며 나무들을 흔적 없이 지워내는 광활한 풍경.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "omniscient high-angle long view from far beyond the ridge, muted early morning daylight contrasting with an intense pale beam",
            "t2i_prompt": "Photorealistic cinematic still. Seen from high above and far beyond the ridge, tilted slightly downward, a thick overcast layer of morning clouds occupies the upper frame while a broad, dense green forest spreads across the lower section. A massive, perfectly cylindrical pale beam of light spans vertically from the greyish textured ceiling of clouds directly into the landscape below, remaining narrow against the vast breadth of the forest. At its contact point in the woods, a circular patch of trees is completely absent, leaving a stark, empty void without cut or burned remnants. Subdued early morning exterior daylight casts a restrained, neutral contrast over the wide environment, but the stark, intense pale illumination of the column washes out the immediate surrounding area, sharply interrupting the muted earthy tones and subdued greens of the untouched woods.",
            "outfit_assignments": [],
            "source_facts": [
              "우거진 숲 한가운데로 수직으로 꽂혀 내린 거대하고 창백한 원통형 빛기둥.",
              "백 미터 앞의 나무를 소리 없이 지운다. 잘리거나 타지 않는다. 존재하던 공간만 텅 빈다.",
              "From high above and far beyond the ridge, a static, slightly oblique downward long view holds the cloud layer near the upper frame, the broad forest below, and the complete vertical reach of the pale cylindrical light column between them.",
              "The column remains narrow against the breadth of the forest while the emptied section at its contact point is readable in the same composition."
            ],
            "visual_inferences": [
              "early_morning_daylight (fixed_element): The early morning timeframe provides soft, ambient illumination across wide landscape views.",
              "pale_light_column (fixed_element): The column is perfectly geometrical, contrasting sharply with the organic shapes of the forest.",
              "cloudy_sky_layer (fixed_element): The heavy cloud layer diffuses the morning sunlight, contributing to the restrained contrast."
            ],
            "creative_decisions": [
              "Render the column with crisp, sharp edges to emphasize its synthetic and absolute destructive power rather than a diffuse, natural beam.",
              "Maintain a consistent desaturated and pale atmospheric tone to unify the varying scales of the shots.",
              "Use the gloomy overcast sky to heighten tension and emphasize the dramatic visual intrusion of the descending pale light weapon."
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [],
            "reference_phrase_kinds": [],
            "owned_object_usage": [],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "5fe4c395e8775587",
              "owned_hash": "e3b0c44298fc1c14",
              "camera_direction_hash": "5ab65c410ba8c2b1",
              "owned_usage_hash": "4f53cda18c2baa0c",
              "validator": "scene_detail_owned_objects.v1",
              "violations": []
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": 9,
        "dependency_reason": "같은 씬의 앞선 샷(Shot 3)에서 확립된 숲속 능선의 이른 아침 빛과 안개 낀 대기 톤(atmosphere_reference)을 공유하며, 빛기둥이라는 새로운 환경 요소를 동일한 시간대의 광활한 풍경 속에 도입.",
        "_shot_index": 5,
        "visible_entities": [],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 9,
            "shot_index": 5
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "wide",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "From high above and far beyond the ridge, a static, slightly oblique downward long view holds the cloud layer near the upper frame, the broad forest below, and the complete vertical reach of the pale cylindrical light column between them. The column remains narrow against the breadth of the forest while the emptied section at its contact point is readable in the same composition; the single emphasized shift from the preceding screen image is the vast increase in camera distance.",
            "lighting_mood": "Daylight over the forest is interrupted by the explicitly pale light column, preserving subdued greens and neutral contrast around its brighter vertical form.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "not_applicable": true,
                "not_applicable_reason": "이 샷의 visible_entities 에 인물이 0명이라 두 인물 사이의 공유 anchor 규칙은 성립하지 않는다",
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=wide — close 규칙은 이 샷에 해당하지 않는다"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": null,
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [],
            "allowed_outlook_pairs": [],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "not_applicable",
            "bg_id": null,
            "reference_usage": "none",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "no chain_bg is bound to this shot — describe the location freely without referencing a background image"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "early_morning_daylight",
                "element_type": "environment_state",
                "character_name": "",
                "description": "Subdued early morning exterior daylight casting a restrained, neutral contrast over the environment. This natural lighting preserves the muted earthy tones of the surroundings without introducing harsh directional shadows.",
                "element_scope": "full",
                "applies_to_shots": [
                  3,
                  5,
                  13
                ],
                "source_facts": [
                  "Exterior daylight is balanced against the active display (Shot 3)",
                  "Daylight over the forest... preserving subdued greens and neutral contrast (Shot 5)",
                  "Exterior daylight retains rugged, restrained contrast (Shot 13)"
                ],
                "visual_inferences": [
                  "The early morning timeframe (05:20-06:00) provides soft, ambient illumination across both macro surface details and wide landscape views."
                ],
                "creative_decisions": [
                  "Maintain a consistent desaturated and pale atmospheric tone to unify the varying scales of the shots."
                ],
                "confidence": "high"
              },
              {
                "element_id": "pale_light_column",
                "element_type": "environment_state",
                "character_name": "",
                "description": "A massive, cylindrical pale beam of light descending vertically into the landscape. It acts as a stark, intense, and unnatural source of pale illumination that visibly impacts the ambient light of the environment it approaches.",
                "element_scope": "full",
                "applies_to_shots": [
                  5,
                  13
                ],
                "source_facts": [
                  "complete vertical reach of the pale cylindrical light column (Shot 5)",
                  "approaching pale column adding only the illumination explicitly present in the scene (Shot 13)"
                ],
                "visual_inferences": [
                  "The column is perfectly geometrical, contrasting sharply with the organic shapes of the forest and ridge.",
                  "Its intense brightness washes out the immediate surrounding area with a harsh, pale glow."
                ],
                "creative_decisions": [
                  "Render the column with crisp, sharp edges to emphasize its synthetic and absolute destructive power rather than a diffuse, natural beam."
                ],
                "confidence": "high"
              },
              {
                "element_id": "cloudy_sky_layer",
                "element_type": "environment_state",
                "character_name": "",
                "description": "A thick, overcast layer of morning clouds occupying the sky above the forest and the ridge. The cloud cover provides a greyish, textured ceiling from which the pale light column descends.",
                "element_scope": "full",
                "applies_to_shots": [
                  5,
                  13
                ],
                "source_facts": [
                  "holds the cloud layer near the upper frame (Shot 5)",
                  "절벽 상공의 구름 낀 하늘 (Shot 13 environment elements)"
                ],
                "visual_inferences": [
                  "The heavy cloud layer diffuses the morning sunlight, directly contributing to the restrained contrast and muted tones of the scene's daylight."
                ],
                "creative_decisions": [
                  "Use the gloomy overcast sky to heighten tension and emphasize the dramatic visual intrusion of the descending pale light weapon."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 9,
                "shot_index": 3,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [],
            "forbidden_refs": [],
            "readiness_policy": "not_applicable",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "9654303216580185"
      },
      {
        "scene_index": 9,
        "heading": "EXT. 숲속 능선 — 낮",
        "beat_title": "절벽 밖으로의 도약과 회피",
        "representative_moment": "윌마가 토니의 하네스를 잡고 절벽 끝에서 허공으로 몸을 던지는 순간.",
        "t2i_variations": [
          {
            "variant_label": "var_1",
            "camera_effect": "Tracking wide shot from below hip height, muted early morning daylight contrasting with intense pale column",
            "t2i_prompt": "Photorealistic cinematic still. The cliff edge anchors the lower-left midground, while C01O01 in dark rescue workwear, a Caucasian man in his 30s, occupies the middle-center midground moving toward the open space beyond the cliff, and C02O02 in a green aerial recon uniform, a Caucasian woman in her 20s, occupies the middle-right midground moving toward the open space beyond the cliff. Seen from below hip height tracking just outside the cliff edge, the two full bodies are suspended in a mid-air leap. C02O02 is in right profile, her trailing foot just separating from the rocky ledge, one hand gripping C01O01's chest harness, gazing into the distance. C01O01 is at three-quarter angle from the left, his torso pulled past the ledge into the void, looking down toward the forest visible far below. The link connecting her belt to his chest harness is pulled taut between their bodies. In the background, a massive, cylindrical pale beam of light descends vertically from a thick, overcast layer of morning clouds, its intense stark glow bleeding into the subdued, restrained early morning exterior daylight over the landscape.",
            "outfit_assignments": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              },
              {
                "character_id": "C02",
                "outlook_id": "O02"
              }
            ],
            "source_facts": [
              "Wilma grabs Tony's harness and throws herself off the cliff",
              "They do not fall but soar long over the forest",
              "The pale light column approaches their previous position"
            ],
            "visual_inferences": [
              "The physical link connects Wilma's belt to Tony's chest harness",
              "Low tracking angle from below hip height captures both full bodies launching into the void, with the ledge in the lower frame",
              "The cloudy sky and early morning daylight provide a restrained backdrop for the intense pale column"
            ],
            "creative_decisions": [
              "Emphasize the taut physical connection between the two characters in mid-air",
              "Use the wide framing from a low angle to capture the vastness of the void and the looming threat in the background"
            ],
            "confidence": "high",
            "applied_frame_spatial_constraint_ids": [
              "fsc_001",
              "fsc_002",
              "fsc_003"
            ],
            "reference_phrase_kinds": [
              "character"
            ],
            "owned_object_usage": [],
            "owned_validation": {
              "schema_version": 2,
              "t2i_prompt_hash": "54bf302e1a849760",
              "owned_hash": "e3b0c44298fc1c14",
              "camera_direction_hash": "480b297baaab2590",
              "owned_usage_hash": "4f53cda18c2baa0c",
              "validator": "scene_detail_owned_objects.v1",
              "violations": []
            }
          }
        ],
        "scene_type": "normal",
        "dependent_scene_index": 9,
        "dependency_reason": "Same environment and identical background setting as S9_Shot5, maintaining the continuity of the overcast sky, forest landscape, and the descending pale light column.",
        "_shot_index": 13,
        "visible_entities": [
          "C01",
          "C02",
          "L05",
          "L06"
        ],
        "render_prompt_card": {
          "schema_version": 1,
          "shot_key": {
            "scene_index": 9,
            "shot_index": 13
          },
          "render_strategy": {
            "mode": "direct",
            "primary_subject": "",
            "framing_scale": "wide",
            "moment_lock": "single still moment from the selected shot only",
            "camera_direction": "Tracking below hip height just outside and parallel to the cliff edge, the camera looks diagonally across the launch path as Wilma and Tony move from the left side of the ledge into open space at frame right. Wilma grips Tony's chest harness as her trailing foot separates from the rock, while Tony's torso is pulled past the edge; both full bodies remain readable and the tightened lateral spacing emphasizes only their changed position beyond the cliff.",
            "lighting_mood": "Exterior daylight retains rugged, restrained contrast, with the approaching pale column adding only the illumination explicitly present in the scene.",
            "perception_mode": "direct",
            "spatial_consistency": {
              "camera_frame_rule": {
                "applies_when": "camera_direction expresses a vertical camera height or viewpoint position and the scene description includes frame-edge surfaces",
                "consistency_check_summary": "elements described inside the frame must be at vertical/depth positions consistent with the camera's height — low cameras place feet/lower-body in the lower frame, overhead cameras place head/shoulders in upper frame, hip-height cameras place torso in frame with head/feet implied beyond edges",
                "core_principles": [
                  "low cameras at ground/quay/floor level place feet and lower legs in the lower frame; chest-height frame-edge surfaces behind the subject are geometrically forbidden",
                  "overhead high angles from above place the top of head and shoulders in the upper-center frame; floor surfaces visible below feet at the lower frame edge are forbidden",
                  "hip-height side cameras place torso and arms in frame; head and feet are implied beyond frame edges, not both fully visible together",
                  "full body shots require either eye-level or high-angle framing where head and feet both fit; low cameras + full silhouette in one frame is forbidden"
                ],
                "forbidden_combinations": [
                  "low camera at ground/quay/floor level + frame-edge surface described at chest/waist height behind the subject",
                  "overhead high angle + feet visible at the lower frame edge with floor visible below them",
                  "hip-height side camera + full silhouette including head and feet visible in same frame",
                  "low angle looking up + ceiling and floor both fully visible in same frame"
                ],
                "recommended_phrasings": [
                  "low observer at ground level, shoes and lower legs occupying the lower frame, upper body extending toward the upper edge",
                  "overhead high angle from above, the top of head and shoulders fill the upper-center frame, the floor visible around them",
                  "at hip height beside the subject, torso and arms in frame, head and lower legs implied beyond frame edges",
                  "low angle looking up at the figure, ceiling lamps in the upper frame, the wall behind extending downward"
                ],
                "self_check_steps": [
                  "identify the camera position phrasing (height + angle)",
                  "list elements described inside the frame (foreground prop, body parts, background surfaces)",
                  "verify each element's vertical/depth placement is geometrically consistent with that camera position",
                  "if mismatch found, adjust frame-edge description to match camera position"
                ],
                "rationale_summary": "T2I 모델은 카메라 위치와 frame 안 요소 vertical 좌표 모순 시 절충 합성 — 인물이 가구에 박히거나 신체 일부가 잘리거나 prop 이 떠 있는 부자연스러운 결과"
              },
              "fg_bg_shared_anchor_rule": {
                "applies_when": "two characters share an interaction (dialogue / handing object / facing / co-watching) AND are described as foreground/background separated AND a shared surface or spatial anchor exists between them (bench / table / door / room)",
                "shared_anchor_required": "describe the shared surface or spatial anchor explicitly so both figures occupy the same physical space",
                "forbidden_phrasings": [
                  "fg/bg separation without naming the shared surface/space",
                  "isolated foreground hand reaching with another figure 'in the rear background' (no shared anchor named)",
                  "one figure 'in the foreground' and another 'in the back' without naming the room/surface they share"
                ],
                "recommended_keywords": [
                  "shared bench / shared table / shared surface",
                  "the same X they both occupy (X = bench, sofa, table, doorway, room)",
                  "across the X from each other",
                  "<prop> crosses the air between them (interaction prop visually links the two)",
                  "the <surface> sharing a single line of sight from the camera"
                ],
                "self_check_steps": [
                  "do the two characters share a single interaction (dialogue, handing, facing, co-watching)?",
                  "is one foreground and the other background-separated?",
                  "is the shared surface/space anchor explicitly named?",
                  "if not, add one of the recommended keywords"
                ],
                "rationale_summary": "fg/bg 분리만 묘사하면 모델이 두 인물을 다른 공간에 배치하거나 인터랙션이 끊긴 어색한 합성 — 공유 anchor 명시가 같은 물리적 공간 강제"
              },
              "primary_framing_rule": {
                "applies_when": "framing_scale ∈ {close, medium, wide, insert} from `shot_staging.framing_scale` enum (render_strategy.framing_scale, production SOT). LLM consumes the enum only; keyword classification is forbidden.",
                "close_framing_rules": {
                  "not_applicable": true,
                  "not_applicable_reason": "framing_scale=wide — close 규칙은 이 샷에 해당하지 않는다"
                },
                "wide_medium_rules": {
                  "primary_and_secondary_both_visible": "primary subject + secondary entities all renderable",
                  "body_part_close_up_forbidden": "body-part close-up of any character is forbidden in wide/medium framing — close-up belongs in a separate shot",
                  "fg_bg_separation_requires_shared_anchor": "if two characters are rendered fg/bg separated, also apply fg_bg_shared_anchor_rule",
                  "subject_reference_policy_cross_ref": "see id_policy.subject_reference_policy (subject_reference_policy SOT — policy enum 에 따라 ID/outlook 사용)"
                },
                "self_check_steps": [
                  "read framing_scale enum from render_strategy.framing_scale (do NOT infer from camera_direction)",
                  "if close: verify only primary subject is fully visible; other entities are partial / soft bg / absent",
                  "verify same character's full body + body-part close-up are not combined",
                  "if close: verify two faces are not both rendered sharp"
                ],
                "rationale_summary": "한 shot 안 framing scale 가 일관되어야 모델이 단일 카메라 셔터로 해석. close/wide/medium 혼합은 multi-camera 합성으로 인식되어 인물 분열 / 부위 분리 / 시점 혼합 합성"
              },
              "rationale_summary": "Rule F/G/J 의 spatial consistency 룰 통합 — camera 위치 vs frame edge 모순, fg/bg 분리 시 공유 anchor 누락, framing scale 혼합. 3 sub-rule 모두 single t2i_prompt = single camera shutter 원칙의 spatial 측면."
            },
            "frame_spatial_contract": {
              "reason": "movement_direction",
              "constraints": [
                {
                  "constraint_id": "fsc_001",
                  "target_kind": "background",
                  "target_id": "",
                  "label": "cliff edge",
                  "screen_zone": "lower_left",
                  "depth_plane": "midground",
                  "gesture_action": "none",
                  "gesture_target_label": ""
                },
                {
                  "constraint_id": "fsc_002",
                  "target_kind": "character",
                  "target_id": "C01",
                  "label": "토니(앤서니 로저스)",
                  "screen_zone": "middle_center",
                  "depth_plane": "midground",
                  "gesture_action": "moves_toward",
                  "gesture_target_label": "open space beyond the cliff"
                },
                {
                  "constraint_id": "fsc_003",
                  "target_kind": "character",
                  "target_id": "C02",
                  "label": "윌마 디어링",
                  "screen_zone": "middle_right",
                  "depth_plane": "midground",
                  "gesture_action": "moves_toward",
                  "gesture_target_label": "open space beyond the cliff"
                }
              ]
            },
            "constraints": [
              "one prompt captures one still moment",
              "do not combine full-body framing and body-part close-up in one frame",
              "camera position and frame-edge elements must be geometrically consistent — see render_strategy.spatial_consistency.camera_frame_rule",
              "two characters in fg/bg separation must share a named surface/space anchor — see render_strategy.spatial_consistency.fg_bg_shared_anchor_rule",
              "framing_scale binds entity rendering — close = primary subject only fully visible, wide/medium = no body-part close-up — see render_strategy.spatial_consistency.primary_framing_rule"
            ]
          },
          "id_policy": {
            "allowed_base_entity_ids": [
              "C01",
              "C02",
              "L05",
              "L06"
            ],
            "allowed_outlook_pairs": [
              {
                "character_id": "C01",
                "outlook_id": "O01"
              },
              {
                "character_id": "C02",
                "outlook_id": "O02"
              }
            ],
            "must_use_composite_character_ids": true,
            "common_noun_required_when": [
              "a reproduced face inside any declared reproduction surface",
              "unregistered extra"
            ],
            "id_use_required_when": [
              "a physical photo/poster/document/picture-frame/map/key prop appearing in the scene — use its P## ID and preserve canonical content from entity_canon"
            ],
            "demographic_fallback_required": true,
            "face_identifiability_rule": {
              "use_entity_id_when": [
                "face identifiable: front, profile, three-quarter, or eyes closed",
                "OTS framing with any visible face"
              ],
              "common_noun_required_when": [
                "back to camera with no visible face",
                "silhouette or blurred outline only",
                "OTS with only back of head/shoulder, face entirely hidden"
              ],
              "id_use_summary": "C##/C##O## allowed only when face is identifiable AND framing is not dominated by an isolated body part",
              "rationale_summary": "C## 사용 시 얼굴 ref 가 자동 결합 — 얼굴이 안 보이는 인물에 C## 부착하면 ref 와 본문 묘사가 충돌"
            },
            "reproduction_surface_rule": {
              "applies": false,
              "id_use": "forbidden — generic descriptor with demographic only",
              "rationale_summary": "C##O## 은 얼굴 reference 이미지를 원본 해상도로 inject 하므로 사진/화면/반사 표면 안 얼굴이 표면 밖 실물 크기로 합성됨. 이 rule 은 표면 안 얼굴 (face) 에만 적용 — 물리적 사진/액자/문서 prop 자체는 P## ID 보존 (Patch A)."
            },
            "demographic_descriptor_policy": {
              "required_on_first_appearance": true,
              "applies_to_id_forms": [
                "C##",
                "C##O##"
              ],
              "token_count_range": [
                1,
                2
              ],
              "components": {
                "ethnicity": [
                  "Asian",
                  "East Asian",
                  "South Asian",
                  "Southeast Asian",
                  "Black",
                  "Middle Eastern",
                  "Hispanic",
                  "Latina",
                  "Latino",
                  "Caucasian"
                ],
                "gender": [
                  "man",
                  "woman",
                  "figure"
                ],
                "age_band": [
                  "in his/her 20s",
                  "in his/her 30s",
                  "in his/her 40s",
                  "young",
                  "middle-aged",
                  "elderly"
                ],
                "role_hint_from_outfit": [
                  "in worker uniform",
                  "in fisher workwear",
                  "in detective coat",
                  "in business suit"
                ]
              },
              "format_template_a": "C##O## in <옷 1-3 단어>, <demographic descriptor>",
              "format_template_b": "C##O##, <demographic descriptor>, <자세 표현>",
              "source_priority": [
                "visual_world_rules.region",
                "visual_world_rules.era",
                "shot context outfit hint"
              ],
              "scenario_dependency_ban": "no work-specific proper nouns (character names, place names) — common nouns only",
              "regional_consistency": "한 작품 내 region 표기는 일관 — 선택된 세분 ethnicity (East Asian / Southeast Asian / South Asian / Caucasian / Black / Hispanic / Latina|o / Middle Eastern) 를 작품 단위 고정"
            },
            "subject_reference_policy": [],
            "constraints": [
              "ID format for visible subjects (C## bare, C##O## composite IDs, or common-noun descriptor) is determined per-subject by the id_policy per-subject policy array — see constraints[2] below",
              "do not use C##O## for a reproduced face inside any element flagged by id_policy.reproduction_surface_rule.applies (declared reproduction surface) — use a generic descriptor with demographic instead. NOTE: the physical photo/poster/document/picture-frame/map/key prop itself MUST use its P## ID — the generic-descriptor rule applies only to a face reproduction, not to the prop entity.",
              "per-subject identity reference policy is provided in id_policy.subject_reference_policy[] — apply each subject's policy when emitting references: (a) id_and_outlook_required ⇒ subject must use C##O## composite ID; (b) base_id_required ⇒ subject must use bare C## only (no outlook); (c) generic_descriptor_allowed ⇒ subject must NOT use C## or C##O##, render with a common/demographic descriptor only. omitted subjects default to id_and_outlook_required",
              "when an ID is required for a subject (policy = id_and_outlook_required or base_id_required) and a C##O## (or C##) is introduced for the first time in this t2i_prompt, append a 1-2 token demographic descriptor (ethnicity + age band, or ethnicity + role hint) consistent with visual_world_rules.region — never use scenario-specific proper nouns. does not apply to generic_descriptor_allowed subjects (no ID emitted)",
              "when an ID is required (policy = id_and_outlook_required or base_id_required), use the format `C##O## in <옷 1-3 단어>, <demographic descriptor>` or `C##O##, <demographic descriptor>, <자세 표현>` — main characters with face refs may include this descriptor too (image inject takes precedence). does not apply to generic_descriptor_allowed subjects — those use a common/demographic descriptor with no C##/C##O##"
            ]
          },
          "background_binding": {
            "mode": "not_applicable",
            "bg_id": null,
            "reference_usage": "none",
            "owned_objects": [],
            "camera_reference": null,
            "close_framing_skips_background_ref": false,
            "constraints": [
              "no chain_bg is bound to this shot — describe the location freely without referencing a background image"
            ]
          },
          "continuity_elements_used": {
            "fixed_elements": [
              {
                "element_id": "early_morning_daylight",
                "element_type": "environment_state",
                "character_name": "",
                "description": "Subdued early morning exterior daylight casting a restrained, neutral contrast over the environment. This natural lighting preserves the muted earthy tones of the surroundings without introducing harsh directional shadows.",
                "element_scope": "full",
                "applies_to_shots": [
                  3,
                  5,
                  13
                ],
                "source_facts": [
                  "Exterior daylight is balanced against the active display (Shot 3)",
                  "Daylight over the forest... preserving subdued greens and neutral contrast (Shot 5)",
                  "Exterior daylight retains rugged, restrained contrast (Shot 13)"
                ],
                "visual_inferences": [
                  "The early morning timeframe (05:20-06:00) provides soft, ambient illumination across both macro surface details and wide landscape views."
                ],
                "creative_decisions": [
                  "Maintain a consistent desaturated and pale atmospheric tone to unify the varying scales of the shots."
                ],
                "confidence": "high"
              },
              {
                "element_id": "pale_light_column",
                "element_type": "environment_state",
                "character_name": "",
                "description": "A massive, cylindrical pale beam of light descending vertically into the landscape. It acts as a stark, intense, and unnatural source of pale illumination that visibly impacts the ambient light of the environment it approaches.",
                "element_scope": "full",
                "applies_to_shots": [
                  5,
                  13
                ],
                "source_facts": [
                  "complete vertical reach of the pale cylindrical light column (Shot 5)",
                  "approaching pale column adding only the illumination explicitly present in the scene (Shot 13)"
                ],
                "visual_inferences": [
                  "The column is perfectly geometrical, contrasting sharply with the organic shapes of the forest and ridge.",
                  "Its intense brightness washes out the immediate surrounding area with a harsh, pale glow."
                ],
                "creative_decisions": [
                  "Render the column with crisp, sharp edges to emphasize its synthetic and absolute destructive power rather than a diffuse, natural beam."
                ],
                "confidence": "high"
              },
              {
                "element_id": "cloudy_sky_layer",
                "element_type": "environment_state",
                "character_name": "",
                "description": "A thick, overcast layer of morning clouds occupying the sky above the forest and the ridge. The cloud cover provides a greyish, textured ceiling from which the pale light column descends.",
                "element_scope": "full",
                "applies_to_shots": [
                  5,
                  13
                ],
                "source_facts": [
                  "holds the cloud layer near the upper frame (Shot 5)",
                  "절벽 상공의 구름 낀 하늘 (Shot 13 environment elements)"
                ],
                "visual_inferences": [
                  "The heavy cloud layer diffuses the morning sunlight, directly contributing to the restrained contrast and muted tones of the scene's daylight."
                ],
                "creative_decisions": [
                  "Use the gloomy overcast sky to heighten tension and emphasize the dramatic visual intrusion of the descending pale light weapon."
                ],
                "confidence": "high"
              }
            ],
            "previous_shot_refs": [
              {
                "scene_index": 9,
                "shot_index": 5,
                "ref_usage": "",
                "kind": "location"
              }
            ],
            "forward_zoom_targets": [],
            "cross_shot_id_substitution_rule": {
              "applies_when": "fixed_elements[i].character_name maps to a C## allowed in id_policy.allowed_base_entity_ids OR a C##O## in id_policy.allowed_outlook_pairs",
              "substitution": "replace the common-noun person reference inside fixed_elements[i].description (e.g. 'An Asian man' / 'a woman' / 'a figure') with the matched C## or C##O##",
              "double_description_forbidden": "do not describe the same character with both a C##/C##O## and a separate common-noun person reference in the same t2i_prompt — produces phantom extra figures",
              "preserve_pose_unchanged": "fixed_elements[i].description 의 자세/위치/상태 keyword 는 그대로 유지 — 인물 참조 토큰만 교체",
              "no_repeat_after_pose": "이미 t2i_prompt 안에서 같은 인물의 자세를 묘사했다면, 같은 fixed_element description 을 별도 문장으로 반복하지 마라",
              "rationale_summary": "fixed_element description 이 보통명사로 작성되었고 같은 인물의 C## 묘사가 t2i_prompt 에 따로 있으면 모델이 두 인물로 인식해 인원 수가 부풀려진다"
            },
            "ref_usage_constraints": {
              "zoom_in_detail": {
                "scope_summary": "same moment, same space, same primary subject; camera moved/zoomed onto a sub-region only",
                "forbidden_additions": [
                  "new persons not in the prior shot",
                  "new props not in the prior shot",
                  "new background elements not in the prior shot",
                  "new poses, gestures, or expressions not in the prior shot",
                  "different moments of action"
                ],
                "required_phrasings": [
                  "describe only the focused sub-region in concrete terms",
                  "explicitly state the same environment (lighting, background) is preserved",
                  "use continuity tokens such as 'the same figure', 'the same room', 'the same surface'"
                ],
                "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy applies per-subject; policy enum (id_and_outlook_required / base_id_required / generic_descriptor_allowed) determines C##/C##O## use",
                "rationale_summary": "zoom_in_detail = 앞 샷의 특정 영역을 확대한 '같은 사진'. 새 정보 추가 금지 — 현재 프레임이 보여주는 부분만 정확히 묘사"
              },
              "exact_background": {
                "scope_summary": "same room, possibly different moment or angle; background carries through, persons/actions are described fresh for this shot",
                "permitted_additions": [
                  "new persons, poses, expressions, actions for this shot's moment",
                  "different camera angle within the same room"
                ],
                "background_consistency_rule": "background elements established in the prior shot must remain consistent (no contradictory walls, doors, furniture)",
                "ignore_keep_handled_elsewhere": "ignore/keep directives are auto-applied via reference image processing — do not echo them inside t2i_prompt",
                "rationale_summary": "같은 공간, 새 순간 — 인물·동작은 이 샷의 instantaneous moment 로 묘사하되 배경은 prior shot 과 일관"
              },
              "atmosphere_reference": {
                "scope_summary": "different room/floor/angle entirely — only mood/tone is shared with the prior shot",
                "forbidden_imports": [
                  "furniture layout from the prior shot",
                  "wall positions from the prior shot",
                  "specific architectural features from the prior shot"
                ],
                "required_handling": "describe the new space's background directly in this shot; carry through only color tone or lighting mood (e.g. 'cool gray ambient' / 'dim warm pool')",
                "rationale_summary": "atmosphere_reference = 분위기 reference 만, 레이아웃 reference 아님. 가구·벽 배치를 끌어오면 모델이 두 공간을 한 공간으로 합성"
              }
            },
            "view_consistency": {
              "framing_scope_options": [
                "third_person",
                "close_up"
              ],
              "single_camera_rule": "one t2i_prompt = one camera position. third-person full body and close-up of the same character must not be combined in a single prompt.",
              "third_person_handling": "describe persons with full or partial body in frame; single-camera consistency forbids combining a partial close-up and the same character's full body in one prompt",
              "close_up_handling": "an isolated framed area of a single subject must not coexist with that subject's full-body description in the same prompt",
              "focus_on_is_focus_area_not_view_switch": "any framing-intent wording designates the area shown by the single camera position; it does not switch the view",
              "subject_reference_policy_cross_ref": "id_policy.subject_reference_policy — subjects with policy = 'base_id_required' or 'generic_descriptor_allowed' must not use outlook reference (C##O## form); additionally, subjects with policy = 'generic_descriptor_allowed' must not use bare C## either — render with a common/demographic descriptor only",
              "prop_orientation_rule": "props the character holds must face one direction only — toward camera OR toward character, not both. forbidden contradiction example: 'photo angled partly toward the camera' + 'eyes fixed on the photograph' (양방향 모순)",
              "mixing_forbidden": true,
              "rationale_summary": "single t2i_prompt = single camera shutter at 1/1000s. 시점 혼합은 모델이 두 카메라 합성으로 해석해 인물이 두 번 등장 또는 부위 분리"
            },
            "constraints": [
              "fixed_elements are continuity inputs derived from prior shots — treat them as binding context, not as source facts to be invented",
              "when fixed_elements[i].character_name maps to a C## in id_policy.allowed_base_entity_ids or a C##O## in id_policy.allowed_outlook_pairs, replace the common-noun person reference inside fixed_elements[i].description with that C##/C##O## and do not also write a separate common-noun reference for the same character in the same t2i_prompt",
              "previous_shot_refs with ref_usage='zoom_in_detail' must preserve the same moment, the same space, and the same lighting as the prior shot — no new persons, props, background elements, poses, or moments",
              "previous_shot_refs with ref_usage='exact_background' may freely describe this shot's persons, poses, and moment, but must keep background elements consistent with the prior shot's established room",
              "previous_shot_refs with ref_usage='atmosphere_reference' must describe a new space directly — do not import furniture layout or wall positions from the prior shot, only mood/tone",
              "one t2i_prompt = one camera position — third-person full body and close-up of the same character must not be combined in a single shot. any framing-intent wording designates the area shown by the single camera position, not a view switch (id_policy.subject_reference_policy applies per-subject)",
              "props held by characters face one direction only — toward camera OR toward character, never both"
            ]
          },
          "asset_requirements": {
            "required_refs": [
              {
                "kind": "character_outlook",
                "id": "C01O01",
                "policy": "required"
              },
              {
                "kind": "character_outlook",
                "id": "C02O02",
                "policy": "required"
              }
            ],
            "forbidden_refs": [],
            "readiness_policy": "block_if_missing",
            "constraints": [
              "do not imply or describe a reference image that is not listed in required_refs (no phantom references)",
              "if a required character_outlook ref is missing, scene image generation must block before the image stage"
            ]
          },
          "render_contracts": [],
          "_card_metadata": {
            "lift_status": {
              "rule_a_lifted": true,
              "rule_c_lifted": true,
              "rule_e_lifted": false,
              "rule_h_lifted": true,
              "id_policy_composite_lifted": true,
              "id_policy_close_framing_face_lifted": true,
              "id_policy_reproduction_surface_lifted": true,
              "continuity_cross_shot_id_substitution_lifted": true,
              "continuity_ref_usage_constraints_lifted": true,
              "continuity_view_consistency_lifted": true,
              "continuity_constraints_extended": true,
              "spatial_camera_frame_lifted": true,
              "spatial_fg_bg_shared_anchor_lifted": true,
              "spatial_primary_framing_lifted": true,
              "spatial_constraints_extended": true
            },
            "rule_source": {
              "camera_rule": "A",
              "owned_rule": "C",
              "close_skip_rule": "E",
              "demographic_rule": "H",
              "composite_id_rule": "C## 사용 규칙 — 참조 이미지 연동 (v17:32-64)",
              "close_framing_face_rule": "극단 클로즈업 표현 (v17:66-85)",
              "reproduction_surface_rule": "사진·포스터·화면·거울 속 인물 규칙 (v17:133-146)",
              "cross_shot_id_substitution_rule": "v20/scene_detail/system.md#continuity-section",
              "ref_usage_rule": "v20/scene_detail/system.md#continuity-section",
              "view_consistency_rule": "v20/scene_detail/system.md#continuity-section",
              "camera_frame_rule": "v20/scene_detail/system.md#spatial-consistency",
              "fg_bg_shared_anchor_rule": "v20/scene_detail/system.md#spatial-consistency",
              "primary_framing_rule": "v20/scene_detail/system.md#spatial-consistency"
            }
          }
        },
        "render_prompt_card_hash": "38f48e1cd11e459c"
      }
    ],
    "consumed_bg_catalog_hash": "5cf6755ace725cda",
    "consumed_shot_binding_hash": "781ec7f3cb94554f"
  },
  "step_id": "scene_detail",
  "run_id": "92a1b074-4cec-4a7c-912d-351b6914bc26",
  "resolved_model": "gemini-pro",
  "updated_at": "2026-09-04T17:42:12.842372+00:00",
  "project_config_snapshot": {}
}