outdoor_canon_photo
outdoor_canon_map
프로젝트 114a8883 · 64스텝 완주 · 스틸 77/77 · 직행 37 · RCV 0 · 전 분석 GPT-5.6 (Sol/Terra/Luna) · 캐논 2ref+9절 직행
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: state-specific character appearance reference (severely_injured): 한국인 성인 남성, 짧은 짙은색 머리와 평범한 성인 남성의 얼굴 윤곽. At dusk in the dense undergrowth of the forest escape corridor, the character shown in image 3 flees with one shoe missing, a large bite wound on the thigh bleeding, wearing a terrified expression and staring toward the darkness among the trees. A black silhouette is about to rush at and attack him. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as state-variant reference (severely_injured) — match the character's body language for this specific state - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match tha
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: state-specific character appearance reference (severely_injured): 한국인 성인 남성, 짧은 짙은색 머리와 평범한 성인 남성의 얼굴 윤곽. Reference image 4: character appearance reference: 한국인 성인으로 추정되는 사람 형태의 검은 실루엣, 얼굴과 머리 세부가 짙은 그림자에 가려져 있다. At dusk, in the dense undergrowth surrounding the collision area, the character shown in image 4 lands forcefully on the back of the character shown in image 3, driving his upper body toward the undergrowth at the instant of impact. One of the injured man's shoes has come off along the escape passage, and a large bite mark on his thigh is bleeding. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as state-variant reference (severely_injured) — match the character's body language for this specific state - use image 4 as character
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: state-specific character appearance reference (severely_injured): 한국인 성인 여성, 자연스러운 짙은색 머리와 평범한 성인 여성의 얼굴 윤곽. Reference image 4: character appearance reference: 한국인 성인 남성 형상, 짙은색 머리와 날카로운 인상의 얼굴 윤곽. At night, the character shown in image 4 and the character shown in image 3 are entangled at a utility pole in a city back alley, within the utility-pole incident area and beside the pole-side fall ground. The teeth of the character shown in image 4 are embedded in the neck flesh of the character shown in image 3 as he tears at her neck. The character shown in image 3 collapses without strength. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as state-variant reference (severely_injured) — match the character's body language for this specific
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: state-specific character appearance reference (dead): 한국인 성인 여성, 자연스러운 짙은색 머리와 평범한 성인 여성의 얼굴 윤곽. At night, beside the utility pole at the incident point, the character shown in image 3 lies limp on the ground in the pole-side fall area of a city back-alley passage. She has collapsed after a man with red eyes tore flesh from her neck with his teeth. A faint streetlight reaches the alley, leaving the surrounding passage dim. No annotations, codes, letters, numbers, text, signage lettering, arrows, or dashed lines are visible. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as state-variant reference (dead) — match the character's body language for this specific state - do not copy poses or compositions from reference images - do not alter
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only A regional news broadcast image shows the outside of the police line area of the property shown in image 1, at the boundary of a police line beside the forensic examination area and body collection position on the Geumwolsan hiking trail. Korean police officers block off the incident scene with police tape while detectives examine the area and recover the bodies. A Korean detective directs the response among them. The reporter is heard but not seen. No annotations, codes, letters, numbers, text, signage lettering, arrows, or dashed lines appear anywhere. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit m
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 40대 중반 여성, 단정한 타원형 얼굴과 짙은색 중간 길이 머리. At the forensic examination area and body collection position within the Restricted Investigation Area, the character shown in image 3 stands among the detectives and directs the incident scene. At the moment her fingertips point toward one side of the scene, police officers examine the area, maintain a police line, and collect the bodies. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that charact
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only At the outside of the police line, beside the forensic examination area and the body collection position, police examine the scene and collect two bodies that were found one after another and cannot be identified by appearance. Hyesu directs the response among a group of detectives. A regional news broadcast is visible on a screen, showing the establishment of an investigation headquarters and an emergency alert through non-legible broadcast graphics without any readable text, letters, numbers, or signage. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference i
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only An exterior view of a rooftop room situated on the roof of an old multi-family villa in the outskirts of Incheon, shown in daylight. An exterior steel staircase added later to the building rises to the rooftop, with open treads, thin steel-plate stringers, supporting posts, and pipe handrails. The rooftop room exterior structure and the surrounding old villa are clearly visible. No people are present. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts with the reference image, follow the reference
Photorealistic cinematic still. Reference image 1: pre-rendered BACKGROUND environment reference — layout/material/lighting identity At eye level and square to the existing television from the attached background, its active front display is centered and occupies slightly less than two fifths of the frame, with a narrow margin of the same living area visible around the set. The screen shows a readable Korean-language local news report about a series of unexplained deaths, with subtle broadcast-resolution texture; sharp focus emphasizes the report itself. The existing small window remains at the peripheral edge of the room, admitting a thin shaft of natural daylight. Daylight and television luminance provide restrained interior contrast without any additional light source. Generate one image: - use image 1 for this location's wall/floor/ceiling materials, furniture identity, and lighting mood - follow the camera position, framing, and action requested in the prompt text — do NOT copy the camera position or composition of image 1 unless the prompt explicitly asks for it - do NOT copy any people from image 1 - architectural openings (doors, windows, thresholds) must match image 1 — same position, type, and material; do NOT invent additional doors or windows unless the prompt text explicitly asks for them - if the prompt text places the camera inside or at one of these openings,
Photorealistic cinematic still. Reference image 1: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Reference image 2: character appearance reference: 한국인 40대 초반 여성, 부드러운 타원형 얼굴과 단정한 짙은색 머리. Photorealistic cinematic still: the unusually distinct circular wrist mark fills the lower-center foreground; the character from Image 1 in a khaki windbreaker, a Korean woman in her early 20s with an oval face and long, straight black hair, occupies the middle-left foreground at a three-quarter angle from the right, one eye and her cheek line at the frame edge, leaning toward the character from Image 2 with eyes fixed on the marked wrist; the character from Image 2 in homewear and apron, a Korean woman in her early 40s with an oval face and neatly arranged dark hair, occupies the middle-right midground in left profile, her eyes fixed on the character from Image 1. From chest height beside the character from Image 1, the marked side of the character from Image 2's wrist faces the lens, the wrist and circular sign in precise focus, her partial profile softly visible beyond it. The two women share the entrance-to-interior axis, with the entrance side receding behind the character from Image 1 and the interior continuing behind the character from Image 2. Thin natural daylight from the small living-room window provides restrained interior contrast, with no additiona
Photorealistic cinematic still. Reference image 1: BACKGROUND from a previous shot (SAME ROOM) — use as-is Reference image 2: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Reference image 3: character appearance reference: 한국인 40대 초반 여성, 부드러운 타원형 얼굴과 단정한 짙은색 머리. Reference image 4: POSE + PLACEMENT GUIDE for this shot — line-art mannequins registered over a faint underlay of this room (pose/placement reference only, NOT identity, appearance, style or environment) Photorealistic cinematic still, with the front door formed by the existing iron door in the middle-right background; the character from Image 2 in a khaki windbreaker, a Korean woman in her early 20s, occupies the middle-right foreground with her weight set toward the front door; the character from Image 3 in an apron over homewear, a Korean woman in her early 40s, stands in the middle-left background with her eyes fixed on the character from Image 2. From an eye-level position deep inside the same rooftop-room interior supplied by the background reference, the room-facing side of the iron door angles toward the camera and frames the character from Image 2 at the interior threshold. the character from Image 2 is at a three-quarter angle from the left, looking off-screen, her oval face bordered by long, straight black hair. the character from Image 3 is at a three-quarter angle from the rig
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. On the daytime coastal bicycle road, beside the coastal margin between the road and the sea, the character shown in image 3 rides a bicycle with one knee bent and pressing the pedal downward in mid-motion. She wears earphones, takes in the smell of the sea, and appears bright and energetic. A dog is heard barking. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearan
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. The character shown in image 3 rides a bicycle along a continuous coastal bicycle road, shown in side profile as she listens attentively to the sound of a rough dog barking and takes in the smell of the sea. She appears bright and energetic. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts
Photorealistic cinematic still. Reference image 1: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. the character from Image 1 in a khaki windbreaker, a young Korean woman, fills the middle-center midground, nearly frontal with a preserved three-quarter-right facial view and facing the camera, leaning into a fixed stare with her hardened face turned toward the confined dogs and their enclosures, her eyes fixed along the lens axis on the Rottweiler's eyes. Her oval face and long, straight black hair occupy most of the frame. Only a restrained, soft-edged portion of the confined Rottweiler's rear head and low-braced forequarters enters the lower foreground, back to camera, its eyes fixed on the character from Image 1 across the same shelter aisle. The static camera rests at the dog's eye level on their shared eyeline. Sole sharp focus rests on the hard set of her eyes and the uninterrupted visual axis between them. Restrained daytime ambient illumination remains subdued in low-key contrast without another specified light source. Visible within the out-of-focus shelter interior are limited front-facing portions of multiple dog enclosures containing confined dogs, positioned in the direction of her hardened gaze. [: out-of-focus shelter interior, limited front-facing portions of multiple occupied dog enclosures behind her, confined dogs visibly inside, s
Photorealistic cinematic still. Reference image 1: PREVIOUS SHOT (SAME MOMENT, zoomed-in reframing) — continuity base Reference image 2: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Multiple dog enclosures occupy the middle-right background, each containing confined dogs; the character from Image 2 in a khaki windbreaker, a young Korean woman, fills the middle-left foreground with her hardened face turned toward the confined dogs and their enclosures, her eyes fixed in their direction. At three-quarter angle from the right, her shoulders are tightened and her hardened oval face remains sharply focused, framed by long, straight black hair. The enclosure fronts form repeating confinement lines in the direction of her gaze, with the confined dogs visible at lower visual emphasis. Restrained daytime ambient light keeps her face subdued without an additional specified light source, with low-key contrast preserving the enclosure details. Generate one image: - use image 1 as the continuity base for a zoomed-in reframing — same moment, same environment, the camera moved closer - keep the environment, lighting, and spatial layout from image 1; render the subjects and their poses as described in the prompt text (for a true zoom they match image 1) - do NOT duplicate body parts - render the focused region (hand, wrist, face area, object, etc.) enlarged, co
Photorealistic cinematic still. Reference image 1: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. From just behind and above the shoulder of the character shown in image 1, the character shown in image 1 is hunched over a 2020s-style rectangular smartphone, a portable touchscreen device with a black screen and thin bezels, with her back mostly turned away and her gaze lowered toward the phone; both hands remain within the upper and side portions of the frame. A 2020s-style rectangular smartphone, a portable touchscreen device with a black screen and thin bezels, rests at a realistic handheld scale in the central lower frame, its black glass front display facing upward and showing a contemporary unanswered outgoing-call interface with no portrait or location image. The phone screen forms the sharp focal plane beneath her lowered face, while her shoulders and hands are slightly soft. Restrained daytime ambient light keeps the display legible without any additional visible light source; only a subdued, soft peripheral suggestion of the shelter work area surrounds the subject. Generate one image: - use image 1 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each c
Photorealistic cinematic still. No reference images. A Korean adult female supermarket employee occupies the middle-center foreground at eye height, her face at a three-quarter angle from the left and her eyes fixed on the person beside the camera. Her round face, dark neatly arranged hair, red supermarket vest, slightly forward-leaning posture, and parted lips define a quiet reply; concern is visible in her expression. Focus rests on her eyes, mouth, and the conversational eyeline just beside the lens. A narrow strip of merchandise aisle recedes at a shallow angle behind her, with one aisle-facing shelf edge forming a narrow vertical boundary at the frame edge. Neutral supermarket ambient light gives clear, even illumination with restrained contrast. Generate one image: - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts with the reference image, follow the reference image and ignore that wording - only render what the scene description asks for — if only a hand or wrist is described, do NOT add the character's face - CRITICAL: Only render what the scene description explicitly describes. If a character's face is visible in the scene, match it to the
Photorealistic cinematic still. Reference image 1: BACKGROUND/ENVIRONMENT from a previous shot at the same location An adult Korean man with a plain face, short dark hair, and a red supermarket vest fills the frame from the upper chest upward at three-quarter angle from the right, his face in the upper-center, eyes directed off-screen to one side, lips slightly parted. In the midst of stocking, he holds and braces a merchandise box against the shelf with his forearms; only the box's top edge and aisle-facing side occupy a small portion of the lower frame, positioned below and clear of his expression. The aisle-facing side of the merchandise shelf sits beside and behind him, limited to the immediate surroundings. Focus rests on his parted lips and off-screen gaze, with the visible box edge preserving the interrupted stocking context. Neutral supermarket ambient light gives clear, even illumination with controlled contrast. Generate one image: - use ONLY the lighting, color palette, and environment mood from image 1 - do NOT copy characters, people, or their appearances from image 1 - do NOT copy the composition or character poses from image 1 - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if
Photorealistic cinematic still. Reference image 1: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Photorealistic cinematic still: the off-screen direction of home lies beyond the middle-left background, and the character from Image 1 in a khaki windbreaker, a Korean woman in her early 20s, occupies the middle-right foreground at a three-quarter angle from the left, looking toward that off-screen direction. Her anxious face fills the right half of the frame, her gaze passing just left of the lens, with long straight black hair falling naturally around her oval face. Her upper body remains poised beside the bicycle handlebar; the handlebar and a small section of the bicycle frame enter the lower edge without obscuring her face. At her eye level, dusk ambient light and existing streetside neon signs place restrained cool tones and selective colored accents across visible surfaces. The mart entrance recedes softly behind her, and the content-bearing faces of distant neon signs appear obliquely in the streetscape. Focus rests sharply on her anxious expression and urgent off-screen eyeline; the entrance, bicycle, and signs remain secondary and softly defined. Generate one image: - use image 1 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - d
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. At night, the character shown in image 3 rides a bicycle toward home along the coastal bicycle road, with the upper body leaning forward and one knee bent while pedaling. The character is seen at a distance along the road, with the coastal margin between the road and the sea beside it. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - image 3: this character faces AWAY from the camera in this shot — the face must NOT be visible; use image 3 only for hair silhouette, body build, and clothing/outfit continuity; show the back of the head and body, do not turn the cha
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. At night, at the “Coastal Road Travel Zone” area of the property, beside the continuous coastal bicycle road and its homeward coastal-road continuation, the character shown in image 3 hurries home by bicycle, staring straight toward home with a tense expression. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording i
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 성인 남성, 얼굴과 머리가 보이지 않고 신발과 길게 드리운 사람 그림자만 확인되는 형상. At night, the character shown in image 3 approaches the rooftop room’s iron entrance and comes to a stop there. The scene is set beside the water tank, near the rooftop-room approach and the rooftop clothesline. Dried sweet-flag leaves blown from the rooftop platform lie around the area, which also includes the rooftop platform and the iron entrance door. The external staircase leading up to the rooftop is a later-added steel extension attached to the building’s exterior, with open treads, thin steel-plate stringers, supporting posts, and pipe railings. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person'
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only At the “Rooftop Room Exterior” area, an iron entrance door to the rooftop room is slightly ajar. The stranger has removed an advertising leaflet lodged in the door gap and is pressing it into the opening seam with his fingers. At that instant, the loosened locking mechanism releases with a click, and the iron door opens slightly. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts with the reference image, follow the reference image and ignore that wording - only render what the scene description as
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only At night, beside the rooftop room’s exterior reflection window at the real filming property, the shadow of a stranger who has just entered through the iron entrance door after using an advertising leaflet against the locking mechanism looms over 강민숙. In the reflected shadows, the stranger has driven his face into her shoulder as if biting into it while she screams, their forms pressed against the glass. The rooftop is accessed by an exterior steel extension stair with open treads, thin sheet-metal stringers, support posts, and pipe railings. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match tha
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. At the Exterior Steel Stair area, beside the later-added exterior open-tread steel stair leading to the rooftop and main rooftop exterior deck, the character shown in image 3 has hurried up from the communal yard and pauses near the rooftop room’s half-open iron entrance door, reacting to a disgusting odor with a tense, hardened expression. The stair has open treads, thin steel stringers, supporting posts, and pipe railings. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. At the edge of the rooftop, the character shown in image 3 has just hurried up from the communal yard after leaving her bicycle in the rear area. She pauses beside the rooftop room’s approach, reacting to a sudden disgusting smell and looking toward the half-open iron entrance door. The rooftop exterior deck includes the later-added open-tread steel stair attached to the building’s outer wall, with thin steel stringers, support posts, and pipe railings. A water tank stands beside the approach, with a rooftop clothesline nearby. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible i
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. At night, the character shown in image 3 has just rushed up the exterior metal staircase after noticing a suddenly foul smell. She stands on the main rooftop exterior deck, frozen at the sight of the rooftop room’s iron entrance door in an unusual state, with the door half-open beside her. The staircase is an added steel structure attached to the building’s exterior, with open steps, thin steel-plate stringers, support posts, and pipe railings. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter
Photorealistic cinematic still. Reference image 1: pre-rendered BACKGROUND environment reference — layout/material/lighting identity Reference image 2: state-specific character appearance reference (dead): 한국인 40대 초반 여성, 부드러운 타원형 얼굴과 단정한 짙은색 머리. Reference image 3: POSE / SUPPORT GUIDE for the immobilized figure — a line-art mannequin registered over a faint room (pose / contact / support-surface reference only, NOT identity, appearance or environment) Reference image 4: POSE + PLACEMENT GUIDE for this shot — line-art mannequins registered over a faint underlay of this room (pose/placement reference only, NOT identity, appearance, style or environment) In the upper-center background, the reference-supplied opened curtain frames the character from Image 2 in homewear and apron, a Korean woman in her early 40s, seated in the middle-center midground. A static camera at chest height from just behind an off-frame observer presents her unobstructed full figure at a three-quarter angle from the left. Her body is motionless and lifeless, her head dropped forward and clearly separated from the curtain, her gaze directed down. Her oval face and dark, neatly arranged hair remain visible beneath the lowered head. Her shoulder and collarbone bear exposed ragged bite-like wounds and torn flesh. One wrist and hand hang fully visible and limp as part of her motionless body; a complete circula
Photorealistic cinematic still. Reference image 1: PREVIOUS SHOT (SAME MOMENT, zoomed-in reframing) — continuity base Reference image 2: state-specific character appearance reference (dead): 한국인 40대 초반 여성, 부드러운 타원형 얼굴과 단정한 짙은색 머리. Reference image 3: POSE / SUPPORT GUIDE for the immobilized figure — a line-art mannequin registered over a faint room (pose / contact / support-surface reference only, NOT identity, appearance or environment) the character from Image 2, an East Asian woman in her 40s, is represented by her lifeless wrist and hand hanging limp as part of her motionless, dead body in the middle-center foreground. Her slack, empty hand holds no crumpled photograph; it has already been removed. The complete circular mark on her wrist has turned dark crimson and is fully visible on the side facing the camera. The wrist and slack hand occupy most of the frame; a narrow portion of the adjacent arm enters from the upper right, preserving her three-quarter-right body orientation beyond the crop. Focus rests sharply on the complete circular design. The established bedroom is reduced to an indistinct field of restrained interior night darkness at the frame edges, with deep surrounding shadows; the dark-crimson mark is a sharp color accent, not a light source. Generate one image: - use image 1 as the continuity base for a zoomed-in reframing — same moment, same environment, t
Photorealistic cinematic still. Reference image 1: PREVIOUS SHOT (SAME MOMENT, zoomed-in reframing) — continuity base Reference image 2: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. In the middle-left foreground, the character from Image 2 in a khaki windbreaker, a Korean woman in her early 20s with an oval face and long straight black hair, faces the camera in a recoiling pose, her mouth open in a scream and her eyes fixed off-screen; in the upper-right background, a completed painted-like red circle rests dark and non-luminous on the visible wall face above the bed. Her screaming face dominates the frame in sharp focus. In the lower-right background of the bedroom they both occupy, a slightly softened but legible Korean adult woman sits motionless behind the opened curtain at a three-quarter-left angle, her head dropped forward and gaze down, lifeless and dead. Her exposed shoulder and collarbone bear ragged bite-like wounds, torn flesh, and dark blood trails; scattered bloodstains and a substantial dark-red pool lie near her seated body. The bed's side and upper edge remain visible beneath the red circle, aligning the foreground scream, the seated corpse, and the wall mark along one depth axis. Restrained low-key interior night darkness surrounds the figures, with deep shadows and dark crimson blood and markings that emit no light; only the im
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. At the “Rooftop Room Exterior” area, beside the main rooftop exterior deck and the iron entrance door of the rooftop room, the character shown in image 3 is fleeing through the gap of the open iron door while screaming. The later-added exterior open-tread steel stair rises along the building’s outer wall, constructed with thin steel-plate stringers, support posts, and pipe railings. A siren sounds at the same time. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where the
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 40대 중반 여성, 단정한 타원형 얼굴과 짙은색 중간 길이 머리. At night, at the common entrance to the multi-family villa and communal courtyard of the neighborhood alley and villa courtyard area, below the rooftop room and the later-added exterior open-tread steel stair, the character shown in image 3 reacts with stunned horror after hearing that her mother is dead. The exterior stair has open treads, thin steel stringers, support posts, and pipe railings. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 30대 남성, 각진 얼굴과 짧고 단정한 짙은색 머리. At night, at the Exterior Steel Stair area beside the rooftop room exterior, the character shown in image 3 looks upward toward the rooftop room with a tense expression. The rooftop is reached by a later-added exterior open-tread steel stair attached to the building’s outer wall, with thin steel stringers, support posts, and pipe railings. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that cha
Photorealistic cinematic still. Reference image 1: BACKGROUND from a previous shot (SAME ROOM) — use as-is The empty space behind the curtain occupies the middle-center midground. The existing curtain edges of the old rooftop residential space connecting a small living room and several rooms frame the left and right margins, with a small window, sink, dining table, bed, and curtains visible. The exact room layout is preserved, including a partial view of the upper surface of the nearby bed beside the cleared space and enough of the meticulously ordered interior to establish scale. The background shown in image 1 anchors the room without obstructing the view. The scene remains focused on the bodyless space: no body, bloodstains, red mark, disturbance, or visible remnant is present. Restrained nighttime exposure and controlled low-key contrast keep the unnaturally pristine room legible without a distinct light source. Generate one image: - use the background, furniture layout, walls, and lighting from image 1 as-is - do NOT copy any standing/moving people from image 1 - from image 1: keep opened curtain edges framing the cleared space - from image 1: keep partial upper face of the nearby bed - from image 1: ignore Ignore the bloodstains, dark blood pool, and red wall mark. - do not copy poses or compositions from reference images - do not alter character identities where their
Photorealistic cinematic still. Reference image 1: BACKGROUND/ENVIRONMENT from a previous shot at the same location Reference image 2: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. In the middle-left foreground, the character from Image 2 in a khaki windbreaker, a Korean woman in her early 20s, holds a forward-leaning mid-turn at three-quarter angle from the right, her eyes looking off-screen toward the empty space behind the curtain in the middle-right background. At her eye level, the camera rests beside her eyeline; her confused eyes occupy the upper center, sharply defined, with only part of her pale, rigid oval face and long, straight black hair visible. [: A narrow defocused curtain edge beside a completely empty cleared space, with no body and no visible remnant.] Restrained nighttime interior exposure, controlled low-key contrast, subdued surrounding detail, no distinct light source. Generate one image: - use ONLY the lighting, color palette, and environment mood from image 1 - do NOT copy characters, people, or their appearances from image 1 - do NOT copy the composition or character poses from image 1 - use image 2 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible i
Photorealistic cinematic still. Reference image 1: pre-rendered BACKGROUND environment reference — layout/material/lighting identity Reference image 2: object appearance reference: 작은 인형 장식이 외부에 매달린 일상용 백팩, 지퍼 수납부와 두 개의 어깨끈을 가진다. From just inside the same door, the object shown in image 2, the already-present backpack, is positioned at natural scale in the lower-left midground, while the already-present plush toy is positioned at natural scale in the lower-right midground. Both remain in shared sharp focus across the same existing floor. The object shown in image 2 is angled so that its front and side are visible relative to the front-facing plush toy, with a small, readable offset between them. Preserve the referenced rooftop room background as excessively clean, meticulously ordered, and entirely free of visible bloodstains or disturbance. The room-facing edge of the same door borders the rear frame, with only a narrow band of the orderly room surrounding the two objects. Restrained nighttime exposure and controlled low-key contrast keep every visible detail legible without introducing a distinct light source. Focus rests on the precise spacing and angle between the object shown in image 2 and the plush toy, not on either object alone. Generate one image: - use image 1 for this location's wall/floor/ceiling materials, furniture identity, and lighting mood - follow the came
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. At the sea-facing sitting place by the bus stop, beside the coastal land edge facing the sea, the character shown in image 3 sits facing the water and holds a mobile phone in her hand. The phone screen visibly indicates an unanswered call from her mother. She appears confused, uncertain whether she truly saw what she thought she saw, and wonders where her mother is. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/o
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. At the coastal area near the bus stop, away from Hyesu and the sea-facing bus-stop waiting point, the character shown in image 3 runs alone toward the house. Her full body is visible, with one foot pushing off the ground and the other leg extended forward. Hyesu calls after her, telling her to ask her mother to phone Hyesu if she sees her. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Reference image 4: character appearance reference: 한국인 40대 중반 여성, 단정한 타원형 얼굴과 짙은색 중간 길이 머리. At night, by the coastal land edge facing the sea, near the coastal bus-stop waiting point and the departure approach leading away from it, the character shown in image 3 runs off alone toward home. The character shown in image 4 remains behind and calls after her, telling her to have her mother phone Hyesu if she finds her. They are seen from a distant vantage point, as if observed by an unidentified observer. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - image 3: this
Photorealistic cinematic still. Reference image 1: pre-rendered BACKGROUND environment reference — layout/material/lighting identity Reference image 2: character appearance reference: 한국인 40대 중반 여성, 단정한 타원형 얼굴과 짙은색 중간 길이 머리. From an eye-height observer position just inside the office entrance, the character from Image 2 in a black detective jumper, a Korean woman in her mid-40s with an oval face and dark medium-length hair, occupies the center-left from above the knees upward, at three-quarter angle from the right, caught mid-step into the office with one knee advanced toward the interior and her smartphone pressed near her ear. Her gaze rests in the distance off-screen, her mouth set and brow tightened by the unanswered call; focus rests on the foreboding tension across her readable face. The same referenced police office interior extends behind her along the entrance axis, subdued under restrained low-key ambient illumination appropriate to night, with no other person visible. Generate one image: - use image 1 for this location's wall/floor/ceiling materials, furniture identity, and lighting mood - follow the camera position, framing, and action requested in the prompt text — do NOT copy the camera position or composition of image 1 unless the prompt explicitly asks for it - do NOT copy any people from image 1 - architectural openings (doors, windows, thresholds) must matc
Photorealistic cinematic still. Reference image 1: PREVIOUS SHOT (SAME MOMENT, zoomed-in reframing) — continuity base Reference image 2: character appearance reference: 한국인 40대 중반 여성, 단정한 타원형 얼굴과 짙은색 중간 길이 머리. At her eye level, the character from Image 2 in a black detective jumper, a Korean woman in her mid-40s with an oval face and dark medium-length hair, occupies the center-left, leaning forward at a three-quarter angle from the right. Her set expression remains sharply readable, her eyes fixed toward the off-screen detective to the right. Her phone rests low along the bottom edge in one hand, its active screen oriented toward the off-screen detective, with part of a saved contact number obliquely legible to the camera. Focus rests on her firm facial expression and the small numerical evidence cue below it. Restrained low-key nighttime office ambience keeps her face readable, with subdued office depth confined to the frame edges. Generate one image: - use image 1 as the continuity base for a zoomed-in reframing — same moment, same environment, the camera moved closer - keep the environment, lighting, and spatial layout from image 1; render the subjects and their poses as described in the prompt text (for a true zoom they match image 1) - do NOT duplicate body parts - render the focused region (hand, wrist, face area, object, etc.) enlarged, consistent with the frame cont
Photorealistic cinematic still. Reference image 1: BACKGROUND from a previous shot (SAME ROOM) — use as-is Reference image 2: character appearance reference: 한국인 40대 중반 여성, 단정한 타원형 얼굴과 짙은색 중간 길이 머리. Reference image 3: character appearance reference: 한국인 30대 남성, 각진 얼굴과 짧고 단정한 짙은색 머리. Reference image 4: POSE + PLACEMENT GUIDE for this shot — line-art mannequins registered over a faint underlay of this room (pose/placement reference only, NOT identity, appearance, style or environment) the character from Image 2 in a black detective jumper, a Korean woman in her mid-40s, stands in the middle-left midground; the character from Image 3 in a brown leather jacket, a Korean man in his 30s, stands in the middle-right midground. From an eye-level observer position several steps beside their shared axis, both figures are fully visible across the same police office and held in the same depth plane, with broad negative space between them. the character from Image 2 is at a three-quarter angle from the right, her weight set toward the character from Image 3, her eyes fixed on him, her oval face tense beneath dark, medium-length hair. the character from Image 3 is at a three-quarter angle from the left, his weight shifted stiffly back, his eyes fixed on the character from Image 2, his angular face firm beneath short dark hair. The visual focus rests on their readable expressions, reciprocal
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 전신이 짧은 검은 털로 덮인 소형 고양이, 뾰족한 귀와 가는 꼬리를 지닌다. At daybreak, the character shown in image 3 pauses on the wall beside the rooftop room and looks toward the illuminated rooftop room and its exterior structure. A cat’s cry is heard as it crosses the wall. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts with the reference image, follow the reference
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 전신이 짧은 검은 털로 덮인 소형 고양이, 뾰족한 귀와 가는 꼬리를 지닌다. At daybreak, beside the rooftop room and its exterior structure on the Main Rooftop, a black cat is suspended in midair beyond the wall, its body fully extended in a leaping pose. It has just glanced back at the lit rooftop room before clearing the wall. Nearby, the exterior stairway is an added steel structure with open treads, thin steel-plate stringers, support posts, and pipe railings. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - eac
Photorealistic cinematic still. No reference images. The red circle reflected in the bathroom mirror occupies the upper-right background; at middle-center in the mirror, a Korean woman in her early 20s in a khaki windbreaker, with an oval face and long, straight black hair, leans tensely over the sink at a three-quarter angle from the right, her fearful reflected eyes fixed off-axis toward the red circle. Her laterally reversed reflected face fills most of the usable mirror surface, with the circle held above one reflected shoulder. The mirror image is coherent and undistorted, with the mirror frame touching the top and right edges; only the sink's upper edge enters along the bottom. Focus rests on her fearful expression and the spatial tension between her gaze and the anomalous circle. Neutral daytime interior ambience with compressed contrast, without an additional light source or colored illumination. [: Narrow bathroom mirror frame touching the top and right edges, with only the sink’s upper edge visible along the bottom of the composition.] Generate one image: - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts with the reference image, follow th
Photorealistic cinematic still. Reference image 1: pre-rendered BACKGROUND environment reference — layout/material/lighting identity Reference image 2: object appearance reference: 한국인 40대 초반 여성과 한국인 20대 초반 여성이 함께 웃는 모습을 담은 일상적인 컬러 가족사진. The shared mother-and-daughter family photograph, the object shown in image 2, occupies the lower-center foreground, face-up in its existing placement on the dining table supplied by the background reference. Enough of the surrounding tabletop remains visible to establish the print’s physical scale. Within the print, a Korean woman in her early 40s in a faded cardigan faces forward, alive and smiling, with dark hair and her eyes directed forward; the outline of her photographed smiling face is collapsed into downward smears within the print texture. The print boundary, the rest of the object shown in image 2, the existing dining table, and everything outside the reproduced face remain geometrically fixed. The downward deformation within the woman’s photographed face is the focal detail. Neutral daytime interior ambience remains stable around the print, without an added colored hue or additional light source. Generate one image: - use image 1 for this location's wall/floor/ceiling materials, furniture identity, and lighting mood - follow the camera position, framing, and action requested in the prompt text — do NOT copy the camera position or
Photorealistic cinematic still. Reference image 1: PREVIOUS SHOT (SAME MOMENT, zoomed-in reframing) — continuity base Reference image 2: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Photorealistic cinematic still: the character from Image 2 in a khaki windbreaker, a Korean woman in her early 20s, occupies the middle-center foreground, her face positioned center-left. At a three-quarter angle from the left, she is twisting into a forceful self-slap, eyes looking down, her palm pressed hard against her own cheek at the instant of impact. Her cheek is visibly compressed beneath the palm; her frightened, tense expression remains unobscured. The hand enters from the right edge, with the forearm held diagonally across the frame. Her oval face and long, straight black hair remain clearly visible. Static observer perspective at face height, with sharp focus on the compressed cheek, palm contact, and strained facial muscles. Neutral daytime interior illumination rests evenly on visible surfaces without an added colored hue or extra light source; the immediate dining-area interior falls softly out of focus behind her without a competing directional cue. Generate one image: - use image 1 as the continuity base for a zoomed-in reframing — same moment, same environment, the camera moved closer - keep the environment, lighting, and spatial layout from image 1
Photorealistic cinematic still. Reference image 1: BACKGROUND/ENVIRONMENT from a previous shot at the same location Reference image 2: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. the character from Image 2 in a khaki windbreaker, a Korean woman in her early 20s, occupies the center-right at three-quarter angle from the right, shoulders locked mid-turn and eyes fixed on the off-screen employee. Her oval face and long, straight black hair remain clearly visible; her facial muscles are held firm in a visibly tense, restrained set expression. The camera focus rests on the firmness around her eyes and mouth, with the reduced camera distance isolating her reaction. Sparse market circulation space recedes softly behind her as shallow immediate context, without another person in frame. Natural early-morning light maintains subdued contrast and a restrained cool-neutral appearance. Generate one image: - use ONLY the lighting, color palette, and environment mood from image 1 - do NOT copy characters, people, or their appearances from image 1 - do NOT copy the composition or character poses from image 1 - use image 2 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - eac
Photorealistic cinematic still. Reference image 1: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Photorealistic cinematic still, with the market exterior CCTV in the upper-left foreground and the character from Image 1 in a khaki windbreaker, a Korean woman in her early 20s, in the middle-right midground, looking off-screen toward the CCTV coverage area. At three-quarter angle from the left, her chin is raised, her oval face visibly tense and set, facial muscles held firm, long straight black hair falling beside her face. The camera remains at her eye level, with focus locked on her shifted gaze and restrained expression. A narrow side and underside edge of the CCTV unit faces the camera, its viewing axis aligned with the witness position behind her. [L##: Immediate market exterior along the CCTV sightline, witness position aligned behind the character from Image 1, only a narrow frame-edge portion visible under restrained natural early-morning light.] Cool-neutral natural illumination rests evenly on visible surfaces with subdued contrast, no added weather effect or artificial color cast. Generate one image: - use image 1 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in
Photorealistic cinematic still. Reference image 1: PREVIOUS SHOT (SAME MOMENT, zoomed-in reframing) — continuity base Reference image 2: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. From an eye-level observer position just forward of her phone side, the character from Image 2 in a khaki windbreaker, a young Korean woman, fills the frame from the shoulders upward at three-quarter angle from the right, her oval face occupying the center-left. Her long, straight black hair hangs naturally, her facial muscles firm and her expression visibly tense and restrained, her gaze fixed into the distance away from the camera. A compact smartphone is pressed to her right ear, its screen facing inward toward her and away from the camera, with only the handset edge and her hand visible along her right cheek. Focus is locked on her tense, set face and the narrow phone edge. The immediate market exterior remains softly defocused at the frame edges with no other person present. Natural early-morning exterior light creates restrained cool-neutral tones and subdued contrast. Generate one image: - use image 1 as the continuity base for a zoomed-in reframing — same moment, same environment, the camera moved closer - keep the environment, lighting, and spatial layout from image 1; render the subjects and their poses as described in the prompt text (for a true zoom they
Photorealistic cinematic still. No reference images. The surveillance image alone fills the frame with a subtly degraded digital texture and an unusually dark central field. On the left, a Korean woman in her early 40s in a dark jumper appears from the upper torso upward in right profile, alive and upright, her weight angled toward the opposing figure, her gaze fixed on it, her facial outline blurred and weakly identifiable. On the right, a partially visible apparently Korean adult humanoid shadow in a dark cloak appears in left profile, leaning subtly toward the woman, its gaze directed toward her; the human-shaped black outline retains faint shoulder and cloak volume, with its face fully obscured, its head unidentifiable, and no discernible features. Suppressed detail conceals both identities, with focus on the narrow space and opposing line of sight between the two figures. Generate one image: - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts with the reference image, follow the reference image and ignore that wording - only render what the scene description asks for — if only a hand or wrist is described, do NOT add the character's face - CRITIC
Photorealistic cinematic still. Reference image 1: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. the character from Image 1's widened eyes occupy the middle-center foreground and nearly fill the image. the character from Image 1, a Korean woman in her early 20s, faces the camera with her head held in a rigid forward lean toward the unseen CCTV monitor, looking off-screen rather than into the lens. Shallow focus is locked on the widened eyes and the taut skin around them; narrow traces of her oval facial contour and long, straight black hair remain at the frame edges. Restrained indoor ambient light keeps the eyes readable, with firm low-key contrast across the surrounding facial surfaces and no additional visible light source. Generate one image: - use image 1 as character appearance reference — match the person's identity where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts with the reference image, follow the reference image and ignore that wording - only render what the scene description asks for — if only a hand or wrist is described, do NOT add the character's face - CRITICAL: Only render wha
Photorealistic cinematic still. Reference image 1: BACKGROUND/ENVIRONMENT from a previous shot at the same location Reference image 2: character appearance reference: 한국인 40대 중반 여성, 단정한 타원형 얼굴과 짙은색 중간 길이 머리. the character from Image 2 in a black detective jumper, a Korean woman in her mid-40s with an oval face and dark medium-length hair, occupies most of the frame at three-quarter angle from the right, leaning forward in a firm requesting posture, looking off-screen rather than at the camera. Her face sits just off center, her lips slightly parted, with clear space extending along her eyeline toward the unseen security employee. The camera is static at her eye level and close facial distance, slightly displaced toward the CCTV-monitor side. Focus rests sharply on the controlled set of her brow and mouth. At the edge nearest her eyeline, part of an active CCTV viewing station remains secondary, its operating side angled away from the camera. Neutral indoor ambient light rests across visible surfaces in firm low-key contrast, with subdued illumination and no additional visible light source. Generate one image: - use ONLY the lighting, color palette, and environment mood from image 1 - do NOT copy characters, people, or their appearances from image 1 - do NOT copy the composition or character poses from image 1 - use image 2 as character appearance reference — match the person
Photorealistic cinematic still. Reference image 1: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. the character from Image 1 in a khaki windbreaker, a Korean woman in her early 20s with an oval face and long straight black hair, fills the frame from brow to upper chest at a three-quarter angle from the left, her widened eyes looking down toward the lower edge. The static camera sits at her eye level very near her face, with sharp focus on her eyes and restrained detail across the surrounding facial planes. A thin, defocused strip of upward-facing enlarged murder-case photographs rests along the lower frame on the desk's receding work surface, still scattered untidily; only indistinct portions of photographed unidentified adult bodies with missing flesh around the neck, shoulders, and face are visible. Ordinary neutral daytime office ambient light creates subdued contrast across her face, with no conspicuous additional light source. Generate one image: - use image 1 as character appearance reference — match the person's identity where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts with the reference
Photorealistic cinematic still. Reference image 1: pre-rendered BACKGROUND environment reference — layout/material/lighting identity The active monitor from the attached background reference fills the frame on a nearly square camera axis, with only a narrow display edge visible; the central focal plane is the sea-location map and the crisp coordinate line “37.484400 N / 126.072486 E.” The offshore marker sits on the visible map face beside the coordinate text, with “Timestamp 2026-02-18 11:21:08,” “Source: GPS / Network,” and “Status: offline” arranged as secondary tracking data around it. Subtle digital texture and controlled screen luminance preserve crisp legibility against ordinary neutral daytime office ambience with subdued contrast. Generate one image: - use image 1 for this location's wall/floor/ceiling materials, furniture identity, and lighting mood - follow the camera position, framing, and action requested in the prompt text — do NOT copy the camera position or composition of image 1 unless the prompt explicitly asks for it - do NOT copy any people from image 1 - architectural openings (doors, windows, thresholds) must match image 1 — same position, type, and material; do NOT invent additional doors or windows unless the prompt text explicitly asks for them - if the prompt text places the camera inside or at one of these openings, camera-side walls and door/windo
Photorealistic cinematic still. Reference image 1: BACKGROUND from a previous shot (SAME ROOM) — use as-is Reference image 2: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Reference image 3: object appearance reference: 해상 위치의 위도·경도를 손으로 옮겨 적은 작은 종이쪽지, 접어서 보관할 수 있는 크기다. The character shown in image 2 appears only as a hand and the adjacent body in the lower-center foreground at waist height, with the body turned away from the viewer, the head beyond the upper frame, and the gaze directed off-screen. The character’s fingers press the folded object shown in image 3 against the pocket opening, with the folded exterior facing the viewer and the written coordinates concealed inside. The established, temporarily empty police-office work area remains softly defocused behind the character. Neutral daytime office ambience, subdued contrast, no conspicuous new light source, and restrained tonal separation between the object shown in image 3 and the character shown in image 2. Generate one image: - use the background, furniture layout, walls, and lighting from image 1 as-is - do NOT copy any standing/moving people from image 1 - from image 1: keep police office work desk surface - from image 1: keep neutral daytime office illumination - from image 1: ignore Ignore the woman looking down at the desk and the enlarged case photographs scattered on the desk. -
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Outside the police station, on the road at the exterior bicycle riding zone, the character shown in image 3 sits on the bicycle saddle with one knee bent while pressing a pedal. The bicycle had been parked outside the police station before the character shown in image 3 took it and began riding. No signage, lettering, numbers, arrows, dashed lines, or other annotations are visible. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene -
Photorealistic cinematic still. Reference image 1: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. At seated eye level beside the monitor, the character from Image 1 in a khaki windbreaker, a Korean woman in her early 20s with an oval face and long, straight black hair, occupies most of the frame at three-quarter angle from the right, her torso inclined toward the monitor and her eyes fixed off-screen toward the mapped ocean location. Focus rests on her rigid expression and unwavering eyeline. A narrow oblique edge of the monitor occupies the near foreground; its content-bearing surface faces her and reveals only a small portion of the mapped ocean location to the camera. The user-facing side of the PC workstation aligns with her seated posture and recedes beyond the frame. Restrained illumination from the active monitor rests across the visible surfaces of her face, surrounded by subdued, cool, low-key ambient contrast without any conspicuous additional light source. Generate one image: - use image 1 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing w
Photorealistic cinematic still. Reference image 1: BACKGROUND/ENVIRONMENT from a previous shot at the same location Reference image 2: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Reference image 3: object appearance reference: 작은 인형 장식이 외부에 매달린 일상용 백팩, 지퍼 수납부와 두 개의 어깨끈을 가진다. The PC-room entrance is in the upper-center background. The character shown in image 2 stands in the middle-center midground with her back to the viewer, paused mid-step at the threshold with one foot slightly ahead and her gaze directed off-screen beyond the entrance. The character shown in image 3 is held close against her back, with its carried side visible. PC workstation rows line both sides of the clear route, forming receding lines toward the entrance without blocking it. The interior remains subdued under natural ambient illumination, with restrained low-key contrast and no conspicuous light source. Generate one image: - use ONLY the lighting, color palette, and environment mood from image 1 - do NOT copy characters, people, or their appearances from image 1 - do NOT copy the composition or character poses from image 1 - use image 2 as character appearance reference — match the person's identity and outfit where visible in the scene - image 2: this character faces AWAY from the camera in this shot — the face must NOT be visible; use image 2 only for hair silhouette,
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only A coastline lies beneath heavy storm clouds, with the sea roughened by the squall. Heavy rain falls as thunder and lightning roll overhead. Su-ryeong rides a bicycle along the continuous coastal bicycle road beside the sea, focused only on finding her mother. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts with the reference image, follow the reference image and ignore that wording - only render what the scene description asks for — if only a hand or wrist is described, do NOT add the character'
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. At the Coastal Road Travel Zone area of the property, beside the continuous coastal bicycle road, the character shown in image 3 stands facing directly into the pounding rain, her expression set and determined as she thinks only of finding her mother. Thunder and lightning break beneath densely gathered dark clouds. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appear
Photorealistic cinematic still. Reference image 1: pre-rendered BACKGROUND environment reference — layout/material/lighting identity Reference image 2: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. From waist height at the living-area opening, the character from Image 2 in a khaki windbreaker, a young East Asian woman with an oval face and long, straight black hair, stands fully visible slightly left of center in the midground, at a three-quarter angle from the right, alive and upright, locked mid-step with one foot ahead, looking off-screen. Within [L06B04: the same rooftop-room living area at dusk], the existing displaced household items lie in irregular foreground and midground layers around her. The existing open window occupies the opposite half, its opening partly toward the camera with no exterior detail resolved; the existing toppled chairs lie at conflicting angles across that half. The existing footprint trace remains visible on the floor below the character from Image 2 and continues into the room. Focus rests on her locked full-body posture and the floor-level intrusion evidence. Subdued environmental dusk illumination fills the interior with restrained low-key contrast, revealing the existing disorder without any object acting as a light source. Generate one image: - use image 1 for this location's wall/floor/ceiling materials, furni
Photorealistic cinematic still. Reference image 1: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. The red circle and bloodied photograph occupy the middle-left background; the character from Image 1 in a khaki windbreaker, a young Korean woman, occupies the middle-right midground at a three-quarter angle from the left, looking toward the red circle and bloodied photograph. Her frightened oval face and upper shoulders dominate the right half of the frame, her posture leaning back, breath caught, long straight black hair falling beside her face. The observer viewpoint rests just inside the bedroom at chest-to-eye height, slightly off her frontal axis. On the left, the marked bedroom wall angles toward the camera, presenting the full red circle; the old photograph sits at its center with its image-bearing face turned into the room, only its placement and blood visible. Focus favors her frightened expression and preserves the tense eyeline to the nested wall mark. Subdued dusk ambient light fills the immediate bedroom area with restrained low-key contrast; the red marking provides a sharp crimson accent without emitting light. Generate one image: - use image 1 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identitie
Photorealistic cinematic still. Reference image 1: PREVIOUS SHOT (SAME MOMENT, zoomed-in reframing) — continuity base Reference image 2: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. At eye height directly before the character from Image 2, her shocked face occupies the central upper frame, and a bloodied old photograph lies canted through the lower-left foreground. the character from Image 2 in a khaki windbreaker, a Korean woman in her early 20s with an oval face and long, straight black hair, faces the camera with her upper body drawn back and the photograph held raised, her eyes fixed downward on the pictured child rather than at the camera. The photograph's image-bearing face angles toward the camera within her line of sight; its visible image contains an old rural restaurant, a round structure on its roof, and two seated 12-year-old girls. One pictured Korean girl sits at a three-quarter angle from the right beside the other child, looking toward the original photograph camera. A restrained blood stain marks the photograph without emitting light. The immediate wall surface remains soft and partial behind the character from Image 2, bearing the red circle left by the removed photograph. Subdued dusk ambient light provides cold environmental illumination and restrained low-key contrast; neither the wall marking nor the photograph acts as a lig
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only At around sunset, the otherwise empty harbor landing lies beneath dense fog after the rain has stopped. A single boat approaches from the sea toward the harbor, its bow emerging through the fog and facing the landing. No other boats or people are visible. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts with the reference image, follow the reference image and ignore that wording - only render what the scene description asks for — if only a hand or wrist is described, do NOT add the character's fa
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. At the otherwise empty harbor landing and waterline edge, with a single harbor boat arriving, the character shown in image 3 stands facing the boat with a resolute expression. She firmly insists that she will reveal her destination only after being allowed to board, determined to find her mother. The scene takes place at sunset. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that chara
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Reference image 4: character appearance reference: 한국인 20대 중반 남성, 선이 고운 수려한 얼굴과 자연스럽게 정돈된 짙은색 머리. At dusk, in the foggy harbor, the character shown in image 3 and the character shown in image 4 stand beside a single arriving harbor boat at the otherwise empty harbor landing area. The boat is secured by its mooring line. The character shown in image 3 asks the character shown in image 4 to take her out by boat so she can find her mother, and the character shown in image 4 agrees because he has to return anyway. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - use
Photorealistic cinematic still. Reference image 1: BACKGROUND from a previous shot (SAME ROOM) — use as-is Reference image 2: character appearance reference: 한국인 40대 중반 여성, 단정한 타원형 얼굴과 짙은색 중간 길이 머리. From chest height across the room, the character from Image 2 in a black detective jumper, Korean woman in her mid-40s, stands slightly off center with her full body visible, at a three-quarter angle from the left, one foot set ahead in a paused mid-step. Her oval face, dark mid-length hair, and lowered gaze are clearly visible. One hand holds a raised mobile phone; the active recorded-message screen faces her and remains oblique rather than square to the camera, small in the composition yet recognizable as the action focus. The established disordered rooftop-room background surrounds her with irregular foreground and background shapes, preserving the same overall state of disarray. Dusk ambient light reveals her and the room through quiet, restrained low-key contrast, with no additional visible light source. Focus rests on the spatial contrast between her full-body stance and the small raised phone. Generate one image: - use the background, furniture layout, walls, and lighting from image 1 as-is - do NOT copy any standing/moving people from image 1 - from image 1: keep displaced household items scattered across the living area - from image 1: keep open window at the opposite si
Photorealistic cinematic still. Reference image 1: PREVIOUS SHOT (SAME MOMENT, zoomed-in reframing) — continuity base Reference image 2: character appearance reference: 한국인 40대 중반 여성, 단정한 타원형 얼굴과 짙은색 중간 길이 머리. Photorealistic cinematic still with the character from Image 2 in a black detective jumper, a Korean woman in her mid-40s, occupying the middle-right foreground. Her oval face and shoulders fill the frame at three-quarter angle from the right, chin lowered, dark medium-length hair framing her face, brow drawn tight, eyes looking down beneath the lower frame edge. Focus rests sharply on the tightened brow and lowered gaze. Dusk ambient light shapes her visible features with restrained low-key contrast and no additional visible light source. A soft, out-of-focus trace of the disordered rooftop-room interior remains at the frame edge. Generate one image: - use image 1 as the continuity base for a zoomed-in reframing — same moment, same environment, the camera moved closer - keep the environment, lighting, and spatial layout from image 1; render the subjects and their poses as described in the prompt text (for a true zoom they match image 1) - do NOT duplicate body parts - render the focused region (hand, wrist, face area, object, etc.) enlarged, consistent with the frame context of image 1 - from image 1: keep softly defocused disordered rooftop-room interior at the frame
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: object appearance reference: 빛바랜 종이사진으로, 1970~1980년대 시골 식당 앞에 앉은 한국인 12세 여자아이 두 명이 담겨 있다. At the Exposed Deck area of the property, at sunset, an old faded photographic print shows two Korean twelve-year-old girls sitting in front of a rural restaurant from the 1970s or 1980s, with a round circular structure on the roof behind them. The photograph is presented as the image Suriyoung shows Inwoo while asking whether he recognizes the place. No writing or lettering is legible anywhere in the image. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - include the object shown in image 3 - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit mus
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 중반 남성, 선이 고운 수려한 얼굴과 자연스럽게 정돈된 짙은색 머리. At dusk on the exposed deck action area beside the entry to the wheelhouse, the character shown in image 3 reacts as though he recognizes the photograph. His eyes tremble with alarm, and his expression is confused and flustered as he denies knowing where the pictured place is. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity and outfit where visible in the scene - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Reference image 3: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Reference image 4: character appearance reference: 한국인 20대 중반 남성, 선이 고운 수려한 얼굴과 자연스럽게 정돈된 짙은색 머리. At dusk, inside the vessel’s wheelhouse by the deck-to-wheelhouse entry, the character shown in image 4 has abruptly run in from the exposed deck and stretches an arm taut toward the steering wheel. The character shown in image 3 has followed him inside and grips his outstretched arm with tense fingers, stopping him. The sea and deck are visible beyond the wheelhouse enclosure. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - use image 3 as character appearance reference — match the person's identity where visible in the scene - use image 4 as character appearance reference — matc
Photorealistic cinematic still. Reference image 1: pre-rendered BACKGROUND environment reference — layout/material/lighting identity Reference image 2: character appearance reference: 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Reference image 3: character appearance reference: 한국인 20대 중반 남성, 선이 고운 수려한 얼굴과 자연스럽게 정돈된 짙은색 머리. Photorealistic cinematic still: the reference-provided steering wheel occupies the middle-center midground, the character from Image 2 in a khaki windbreaker, an East Asian woman in her early 20s, occupies the middle-right midground, and the character from Image 3 in a green waterproof jumper, an East Asian man in his mid-20s, occupies the middle-left midground. From eye level near the rear of the wheelhouse, both figures are visible from head to thigh within the same compressed steering space. the character from Image 3 is at a three-quarter angle from the right, braced left of center against the wheel, one hand fixed around it, his eyes fixed on the character from Image 2; his refined face, delicate features, and dark hair remain clearly readable. the character from Image 2 is at a three-quarter angle from the left, leaning into her blocking arm from the right, her forearm set across his wheel arm, her eyes fixed on the character from Image 3; her oval face and long, straight black hair remain clearly readable. The steering wheel stays between their oppos
Photorealistic cinematic still. Reference image 1: PREVIOUS SHOT (SAME MOMENT, zoomed-in reframing) — continuity base Reference image 2: character appearance reference: 한국인 20대 중반 남성, 선이 고운 수려한 얼굴과 자연스럽게 정돈된 짙은색 머리. In the middle-center foreground, the character from Image 2 in a green waterproof jumper, a Korean man in his mid-20s, fills most of the frame from a left three-quarter angle, his face extending across center-left. His handsome face has fine, delicate features, dark hair, and a visibly drained, bloodless complexion; his shoulders are locked mid-breath, his eyes fixed directly on the camera. Focus rests on the sudden loss of color across his expression. Subdued ambient sunset illumination rests across his visible features and jumper in restrained low-key contrast, with no additional interior light source. Only a narrow, softly receding strip of the compact fishing-boat wheelhouse interior remains visible beyond his face. Generate one image: - use image 1 as the continuity base for a zoomed-in reframing — same moment, same environment, the camera moved closer - keep the environment, lighting, and spatial layout from image 1; render the subjects and their poses as described in the prompt text (for a true zoom they match image 1) - do NOT duplicate body parts - render the focused region (hand, wrist, face area, object, etc.) enlarged, consistent with the frame contex
Photorealistic cinematic still. Reference image 1: BACKGROUND/ENVIRONMENT from a previous shot at the same location Reference image 2: state-specific character appearance reference (unconscious): 한국인 20대 초반 여성, 매끈한 타원형 얼굴과 곧게 늘어진 긴 검은 머리. Reference image 3: character appearance reference: 한국인 20대 중반 남성, 선이 고운 수려한 얼굴과 자연스럽게 정돈된 짙은색 머리. the character from Image 2 in a khaki windbreaker, a Korean woman in her early 20s, occupies the middle-left foreground; the character from Image 3 in a green waterproof jumper, a Korean man in his mid-20s, occupies the middle-right foreground. At shoulder height beside the pair, a slightly downward observer view isolates the contact point of her forehead against his shoulder. the character from Image 2 is in right profile with eyes closed, unconscious, slack and limp, her straight long black hair and upper body sloping into the shared contact from frame left, her weight concentrated against his shoulder. the character from Image 3 is at three-quarter angle from the left, his shoulder and bracing torso supporting her from frame right, his partially visible face softer than the contact point, eyes fixed on the character from Image 2. The compact wheelhouse interior remains softly visible only along the frame edges. Subdued ambient sunset light rests across her slack face, his supporting shoulder, and their clothing with restrained low-key contras
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only At sea at dusk, In-woo’s small fishing boat appears as a tiny silhouette on the golden reflection of the setting sun. Surrounded by open sea, the boat is gradually entering a dense band of fog ahead, the fog seeming to erase the boundary between the sea and the world. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts with the reference image, follow the reference image and ignore that wording - only render what the scene description asks for — if only a hand or wrist is described, do NOT add the c
Photorealistic cinematic still. Reference image 1: LOCATION PHOTOGRAPH — the real filming property this shot happens at Reference image 2: SITE-PLAN DRAWING of the same property — layout source only Around sunset, In-woo’s small fishing boat slowly enters a dense band of fog ahead, its silhouette faintly visible amid the open sea surrounding it. A golden reflection stretches across the surface of the sea. No people are visible. Generate one image: - image 1 is the SOLE source of how this place looks: building, materials, aging, colours, surroundings - do NOT copy image 1's camera angle, composition, time of day or lighting — the prompt text defines this shot's moment - do NOT copy any people from image 1 - use image 2 ONLY to understand where things are on the property (positions, adjacency, routes) - NEVER draw image 2 itself or its colours, circles, markers or labels into the output - do not copy poses or compositions from reference images - do not alter character identities where their face is visible in the scene - each character's clothing/outfit must match that character's appearance reference image EXACTLY; if any clothing wording in the text conflicts with the reference image, follow the reference image and ignore that wording - only render what the scene description asks for — if only a hand or wrist is described, do NOT add the character's face - CRITICAL: Only ren
직행 스틸 합계: 37 / 77