💃 Клип с толпой клонов в бетонном зале

💃 Клип с толпой клонов в бетонном зале

Промпт / Prompt

REFERENCES @video1 = Strict reference for the original action, character movements, poses, facial expressions, gaze, camera work, editing, timing, and audio. @image1 = Location reference, replacing the original environment. @image2 = Complete appearance reference for the replacement of the original green character. @image3 = THE ONLY appearance reference for the replacement of every original red character, without a single exception. ONE-SENTENCE SUMMARY Recreate the sequence from @video1 in the location shown in @image1, replacing the green character with the person from @image2 and replacing every single red character - foreground, background, occluded, blurred, and briefly visible - with the person from @image3, preserving the original choreography, emotions, crowd counts, and camera structure. GLOBAL SETTINGS Character replacement: Completely replace the original green character with the person from @image2. Replace every original red character with a separate instance of the person from @image3. This includes red characters in the background, characters partially occluded by other bodies, characters visible for a fraction of a frame during transitions, and characters that enter the frame only at the very end of the final pull-out. RED CAST MASTER RULE (highest priority) @image3 is the casting card for the entire red ensemble. Every red figure in every shot, at every timestamp, from every angle, must wear the exact appearance from @image3: same face, same hair, same body type, same clothing, same footwear, same accessories, same colors. Instances of the @image3 person differ only in pose, expression, sweat, subtle lighting and depth-of-field as dictated by @video1 - never in identity or outfit. There is no "close enough" version: there is exactly one red identity in this video, and it is @image3. Green and red are labels for the ORIGINAL characters only. Do not retain those colors unless they exist in the corresponding reference image. After replacement, no red-colored clothing, masks, hoods, or skin remains anywhere in frame unless @image3 or @image2 contains it. Appearance fidelity: Transfer each character reference's entire appearance - face, hair, body, proportions, clothing, footwear, and accessories, including seams, fabric texture, logos, and small details. Leave no remnants of the original identities, masks, hoods, or outfits. Identity consistency across cuts: Keep each replacement's appearance identical through every cut, turn, occlusion, floor contact, and reappearance. When a character is lost behind another body and re-emerges, it must be the same face and the same outfit. Preserve the original number of characters and their individual roles and positions. Crowd count lock: Match the original head count in every shot exactly - never merge two reds into one, never add a stand-in, never leave a gap where a red dancer used to be. If the original has 7-8 reds in the ring, there are 7-8 @image3 instances; if it has ~30 in the final overhead, there are ~30 @image3 instances. Environment replacement: Use @image1 as the authoritative reference for the location's architecture, materials, scenery, palette, and lighting conditions. Completely replace the original setting rather than blending the two environments. Render the referenced location coherently from every camera angle, including the strict top-down angle where only floor material and reflections are visible. Spatial integration: Preserve the original blocking, relative character positions, movement paths, and contact events within the new location. Maintain believable scale, ground contact, shadows, reflections, and occlusion. The mirror-like floor of the original must be replaced with the surface from @image1 while still reflecting the new characters plausibly. Integrate any interaction surfaces or objects required by the original action naturally into the new setting. Motion and performance: Match each original character's movements, poses, facial expressions, gaze, and interactions frame by frame - the soloist's performance and the ensemble's synchronized choreography alike. Preserve action direction, speed, emotional intensity, and synchronization. All red dancers replace at the same instant, in the same frame of their movement, as the originals. Camera and format: Use @video1 as the strict reference for camera movement, rhythm, and subject trajectory. Do not change the camera structure. Preserve framing, perspective, shot order, cuts, duration, and aspect ratio. SHOT-BY-SHOT REPLACEMENT TIMELINE Follow the existing timeline of @video1 exactly, from its first frame to its last. Every shot below requires the red replacement to be already complete on its first frame - no character may "change over" mid-shot. SHOT 1 - 00:00-00:01 - tight ring, close-up / very close-up 7-8 red characters stand shoulder-to-shoulder in a tight semicircle around the green soloist, faces toward her and toward camera, mostly frontal or three-quarter. Replace all 7-8 with @image3 instances. Because they overlap at the shoulders and heads, each partially visible head and shoulder must still read as @image3 - not a generic dancer. Frontal and three-quarter angles, intimate spacing, static poses, silent staring, no physical contact. TRANSITION - 00:01-00:02 - fast push-in / zoom transition Red silhouettes smear past the frame edges for fractions of a second. These blurred figures must still be @image3 in appearance and clothing color - no original red smears remaining at the edges. SHOT 2 - 00:02-00:08 - wide frontal group shot, V-formation 12-14 red dancers in a wedge behind the green soloist, all facing camera, moving in unison: 00:02-00:04 torso tilts, arm swings, bent-knee steps; 00:05-00:07 drop to one knee with fists punched up; 00:08 airborne jumps with sharp leg work. Replace all 12-14 with @image3 instances, including the 4-5 back-row dancers visible only as torsos and heads behind the front line. Front line fully @image3; back row must be visibly the same face and outfit, not a generic silhouette. The camera cranes down and pushes in during this shot - as figures grow larger, their fidelity to @image3 must increase, not degrade. SHOT 3 - 00:08-00:10 - strict top-down (90°) wide shot The full ensemble lies in diagonal parallel rows on the floor, heads aligned, spaced roughly 1.5-2 m apart, writhing and rolling in sync; the green soloist moves among them. Replace every figure - approximately 28-32 - with @image3 instances seen from directly above. Top-down angle must show the same hairline, hair parting, shoulder line, clothing cut and footwear. Distances, row angles and grid pattern stay identical to the original. SHOT 4 - 00:10-00:11 - floor-level tracking medium shot Group A: 3-4 red characters lie on their sides in profile and half-frontal, one behind another along a diagonal behind the crawling green soloist, echoing her serpentine motion. Group B: 5-6 red characters stand behind them, dancing in place; in the original these have their hoods off, showing faces, beards and hair. Replace both groups with @image3 instances. Group B matters most: these are the only open-face red characters in the video, so their faces must be unmistakably @image3 - same facial features, same hair, same beard or clean-shaven state as the reference, no leftover original faces. Group A is seen in profile, so the profile silhouette of @image3's nose, jawline and hairstyle must be preserved. Group B is partly occluded by the lying figures in the foreground; keep that occlusion. SHOT 5 - 00:12-00:16 - top-down, slow pull-out and pan Camera retreats and pans, revealing more and more rows of red dancers writhing on the floor. New @image3 figures gradually slide into frame at the top and bottom edges, creating the illusion of an endless field of people. Every newly revealed dancer is an @image3 instance. By the last frame, the entire visible field - every single body, including those only half in frame at the edges - is @image3. No original red character may be revealed by the pull-out. ANGLE COVERAGE The red replacement must hold from every camera setup in @video1: frontal close-up, three-quarter, profile (lying figures), full-body wide, low floor-level angle, and strict overhead. Each angle preserves the same identity, same hair, same clothing and same footwear - only the perspective changes. AUDIO REQUIREMENTS Preserve the original audio from @video1 unchanged, including dialogue, voices, music, ambient sound, sound effects, and synchronization. Preserve any original lip-sync timing. NEGATIVES No unreplaced red characters at any timestamp, in any shot, at any angle, in any corner of the frame; no original red clothing, hoods, masks or red silhouettes - not even blurred, not even for one frame; no original green character; no original faces, beards or outfits remaining; no identity swapping; no generic or averaged faces in the crowd; no appearance drift between shots or after occlusions; no blending of @image2 and @image3 identities; no extra or missing characters; no merging or doubling of dancers; no remnants of the original location; no altered choreography; no changed expressions; no camera changes; no additional cuts; no retiming; no added dialogue or music; no added subtitles or watermarks. TECHNICAL NOTE Character and location replacement is applied on frame one and held through the final frame. Equal visual weight and detail across all @image3 instances: no "hero" replacement in the foreground with a different-looking crowd in the background.

Модель / Model: Seedance

Тип / Type: видео / video

Автор / Author: Mila Mi

Категории / Categories: Портреты и люди, Арт и иллюстрация

Теги / Tags: портрет

Похожие промпты / Related