🎬 Step 3 — Video Prompt (Seedance 2.5)
Character sheet ready hone ke baad, Seedance 2.5 me 5 images isi order me (image_1…image_5) upload karo, phir ye prompt paste karo.
Video Prompt:
Vertical 9:16 format, continuous cinematic shot in one single take, 23 seconds total, photorealistic real-life, realistic physics, strong spatial continuity, synchronized dialogue.
0.0–3.0s: Begins at the feet of the main character from <<<image_1>>> walking naturally through a busy colonnaded city street inspired by <<<image_2>>> — white pillared arcade, shopfronts, parked cars, autos, street vendors and pedestrians, bright midday daylight, clear blue sky. The camera smoothly tilts up along the body to a front close-up. The character holds <<<image_4>>> against one ear as if on a call, while naturally holding <<<image_5>>> in the other hand. <<<image_3>>> is worn naturally as sunglasses. IMPORTANT: the sunglasses must exactly match <<<image_3>>> and the phone must exactly match <<<image_4>>>, overriding any different eyewear or phone visible in the character reference.
3.0–4.5s: A rushed person approaches head-on carrying folders and loose papers, accidentally bumping the main character. Papers scatter everywhere. The character reacts with surprise, loses balance, and begins falling backward. <<<image_3>>>, <<<image_4>>>, and <<<image_5>>> separate clearly from the character due to the collision, following physically believable trajectories.
4.5–6.0s: Time rapidly slows into a dramatic bullet-time effect. The character keeps falling in extreme slow motion while the scattered papers and released objects hang suspended around them.
6.0–8.5s: The camera moves closer to the suspended <<<image_3>>> (sunglasses), nearly frozen in its natural position, shallow depth of field, the falling character softly visible in the background.
8.5–11.0s: The camera pans across the same frozen scene toward <<<image_4>>> (phone), clearly separated from the character, preserving its exact appearance and turning its most recognizable front (call screen) toward the camera.
11.0–13.5s: The camera pans toward <<<image_5>>> (coffee cup), suspended after being released. In extreme slow motion, the ice, liquid and lid react naturally to the impact with realistic gravity, inertia and material behavior.
13.5–15.5s: The camera pulls back to reveal the full frozen scene: the character falling, papers floating everywhere, and the three reference objects occupying their natural positions in space.
15.5–17.0s: Time abruptly snaps back to normal speed. Gravity and momentum resume simultaneously. The character, papers and released objects drop to the ground; the character lands on the pavement and nearby pedestrians glance over in surprise.
17.0–19.5s: The character sits up on the pavement amid the scattered papers, the phone lying on the ground beside him. He calmly reaches down and picks up the <<<image_5>>> coffee cup (intact) from the ground.
19.5–23.0s: Holding the coffee cup, the character looks directly into the camera lens with a relaxed, confident smile and speaks toward the viewer with natural, accurate lip-sync. He says, in a casual friendly Hinglish tone (clear male voice): "Relax guys, अगर आपको ये prompt चाहिए तो comment करो coffee." A subtle friendly hand gesture. Hold steadily on his upper body and face, centred, with clean empty space in the frame for on-screen text to be added later.
Throughout the sequence, maintain the exact appearance of each reference object. All must originate from the character and separate visibly during collision — never appearing spontaneously. During bullet time, objects remain anchored in space while the camera travels to them for close-ups. No duplicates, no teleportation, and no objects flying into the lens. Do not render any on-screen text or captions. Premium fashion-film cinematography, realistic materials, natural lighting, shallow depth of field, and a practical high-speed photography feel.
Character sheet ready hone ke baad, Seedance 2.5 me 5 images isi order me (image_1…image_5) upload karo, phir ye prompt paste karo.
Video Prompt:
Vertical 9:16 format, continuous cinematic shot in one single take, 23 seconds total, photorealistic real-life, realistic physics, strong spatial continuity, synchronized dialogue.
0.0–3.0s: Begins at the feet of the main character from <<<image_1>>> walking naturally through a busy colonnaded city street inspired by <<<image_2>>> — white pillared arcade, shopfronts, parked cars, autos, street vendors and pedestrians, bright midday daylight, clear blue sky. The camera smoothly tilts up along the body to a front close-up. The character holds <<<image_4>>> against one ear as if on a call, while naturally holding <<<image_5>>> in the other hand. <<<image_3>>> is worn naturally as sunglasses. IMPORTANT: the sunglasses must exactly match <<<image_3>>> and the phone must exactly match <<<image_4>>>, overriding any different eyewear or phone visible in the character reference.
3.0–4.5s: A rushed person approaches head-on carrying folders and loose papers, accidentally bumping the main character. Papers scatter everywhere. The character reacts with surprise, loses balance, and begins falling backward. <<<image_3>>>, <<<image_4>>>, and <<<image_5>>> separate clearly from the character due to the collision, following physically believable trajectories.
4.5–6.0s: Time rapidly slows into a dramatic bullet-time effect. The character keeps falling in extreme slow motion while the scattered papers and released objects hang suspended around them.
6.0–8.5s: The camera moves closer to the suspended <<<image_3>>> (sunglasses), nearly frozen in its natural position, shallow depth of field, the falling character softly visible in the background.
8.5–11.0s: The camera pans across the same frozen scene toward <<<image_4>>> (phone), clearly separated from the character, preserving its exact appearance and turning its most recognizable front (call screen) toward the camera.
11.0–13.5s: The camera pans toward <<<image_5>>> (coffee cup), suspended after being released. In extreme slow motion, the ice, liquid and lid react naturally to the impact with realistic gravity, inertia and material behavior.
13.5–15.5s: The camera pulls back to reveal the full frozen scene: the character falling, papers floating everywhere, and the three reference objects occupying their natural positions in space.
15.5–17.0s: Time abruptly snaps back to normal speed. Gravity and momentum resume simultaneously. The character, papers and released objects drop to the ground; the character lands on the pavement and nearby pedestrians glance over in surprise.
17.0–19.5s: The character sits up on the pavement amid the scattered papers, the phone lying on the ground beside him. He calmly reaches down and picks up the <<<image_5>>> coffee cup (intact) from the ground.
19.5–23.0s: Holding the coffee cup, the character looks directly into the camera lens with a relaxed, confident smile and speaks toward the viewer with natural, accurate lip-sync. He says, in a casual friendly Hinglish tone (clear male voice): "Relax guys, अगर आपको ये prompt चाहिए तो comment करो coffee." A subtle friendly hand gesture. Hold steadily on his upper body and face, centred, with clean empty space in the frame for on-screen text to be added later.
Throughout the sequence, maintain the exact appearance of each reference object. All must originate from the character and separate visibly during collision — never appearing spontaneously. During bullet time, objects remain anchored in space while the camera travels to them for close-ups. No duplicates, no teleportation, and no objects flying into the lens. Do not render any on-screen text or captions. Premium fashion-film cinematography, realistic materials, natural lighting, shallow depth of field, and a practical high-speed photography feel.
Prompt:
Use my uploaded photo as the PRIMARY and ONLY reference for the person. Preserve my exact facial identity and recognizable features — face shape, forehead, hairline, hairstyle, eyebrows, eyes, nose, lips, teeth, jawline, ears, skin tone, facial proportions, and natural expression. Do not redesign, beautify, age, slim, or alter my face. The final portrait must clearly look like the same person from my uploaded photo.
Create an ultra-realistic handcrafted string-art portrait based on my uploaded photo.
The portrait is displayed on a large rectangular white/off-white wooden canvas board, photographed straight-on. The artwork is created using hundreds of small metal pins/nails carefully placed around the outline and facial features, with dark green, black, brown, beige, and skin-tone threads stretched tightly between the pins to form the portrait.
Face & Hair
Recreate my face accurately using dense, intricate thread lines. Use fine skin-tone and light beige threads for the forehead, cheeks, nose, and facial highlights, with darker brown threads creating realistic shadows and facial contours. Use black/dark brown threads for the hair, eyebrows, beard/facial hair if present, and glasses if present. Preserve the exact hairstyle and facial proportions from my reference photo.
The facial details should look like they were genuinely constructed from thousands of carefully layered threads rather than digitally painted.
Clothing
Recreate the clothing visible in my reference photo using dark green thread art, with dense overlapping thread patterns. Preserve the overall clothing shape and neckline from the reference.
String-Art Construction
Make every part visibly handmade:
• Hundreds of small silver/metal pins around the portrait perimeter
• Additional pins defining facial contours and important details
• Threads stretched from pin to pin in different directions
• Dense criss-cross thread patterns
• Layered threads creating realistic gradients, shadows, highlights, and depth
• Slightly imperfect handmade variations in thread placement
• Clearly visible individual threads
• Real physical pins with tiny metallic highlights
• No smooth digital painting appearance
Background & Presentation
Keep the background minimal, clean, matte white/off-white, resembling a real artist’s wooden string-art board. Include a subtle rectangular border/frame around the artwork.
The final image should look like a real photograph of a finished professional string-art portrait hanging/displayed on an artist’s work board, not an illustration or CGI render.
Use soft natural indoor lighting, realistic shadows cast by the raised threads and pins, subtle depth, authentic material texture, high micro-detail, realistic wood/canvas surface, and professional product-photography quality.
Composition: centered head-and-shoulders portrait, straight-on camera angle, symmetrical framing, portrait orientation, full artwork visible.
Style: photorealistic handcrafted string art, intricate nail-and-thread portrait, museum-quality handmade artwork, extremely detailed, tactile physical materials, realistic thread fibers, realistic metallic pins, natural imperfections.
Important: The person’s identity and facial structure must come only from my uploaded reference photo. Do not substitute another face or create a generic person. No extra people, no text, no logos, no watermark, no cartoon effect, no AI-painted skin, no plastic-looking surfaces.
Use my uploaded photo as the PRIMARY and ONLY reference for the person. Preserve my exact facial identity and recognizable features — face shape, forehead, hairline, hairstyle, eyebrows, eyes, nose, lips, teeth, jawline, ears, skin tone, facial proportions, and natural expression. Do not redesign, beautify, age, slim, or alter my face. The final portrait must clearly look like the same person from my uploaded photo.
Create an ultra-realistic handcrafted string-art portrait based on my uploaded photo.
The portrait is displayed on a large rectangular white/off-white wooden canvas board, photographed straight-on. The artwork is created using hundreds of small metal pins/nails carefully placed around the outline and facial features, with dark green, black, brown, beige, and skin-tone threads stretched tightly between the pins to form the portrait.
Face & Hair
Recreate my face accurately using dense, intricate thread lines. Use fine skin-tone and light beige threads for the forehead, cheeks, nose, and facial highlights, with darker brown threads creating realistic shadows and facial contours. Use black/dark brown threads for the hair, eyebrows, beard/facial hair if present, and glasses if present. Preserve the exact hairstyle and facial proportions from my reference photo.
The facial details should look like they were genuinely constructed from thousands of carefully layered threads rather than digitally painted.
Clothing
Recreate the clothing visible in my reference photo using dark green thread art, with dense overlapping thread patterns. Preserve the overall clothing shape and neckline from the reference.
String-Art Construction
Make every part visibly handmade:
• Hundreds of small silver/metal pins around the portrait perimeter
• Additional pins defining facial contours and important details
• Threads stretched from pin to pin in different directions
• Dense criss-cross thread patterns
• Layered threads creating realistic gradients, shadows, highlights, and depth
• Slightly imperfect handmade variations in thread placement
• Clearly visible individual threads
• Real physical pins with tiny metallic highlights
• No smooth digital painting appearance
Background & Presentation
Keep the background minimal, clean, matte white/off-white, resembling a real artist’s wooden string-art board. Include a subtle rectangular border/frame around the artwork.
The final image should look like a real photograph of a finished professional string-art portrait hanging/displayed on an artist’s work board, not an illustration or CGI render.
Use soft natural indoor lighting, realistic shadows cast by the raised threads and pins, subtle depth, authentic material texture, high micro-detail, realistic wood/canvas surface, and professional product-photography quality.
Composition: centered head-and-shoulders portrait, straight-on camera angle, symmetrical framing, portrait orientation, full artwork visible.
Style: photorealistic handcrafted string art, intricate nail-and-thread portrait, museum-quality handmade artwork, extremely detailed, tactile physical materials, realistic thread fibers, realistic metallic pins, natural imperfections.
Important: The person’s identity and facial structure must come only from my uploaded reference photo. Do not substitute another face or create a generic person. No extra people, no text, no logos, no watermark, no cartoon effect, no AI-painted skin, no plastic-looking surfaces.
Video Prompt 2 Use the uploaded reference image as the ONLY and EXACT reference for the final string-art artwork. Preserve the exact woman's pose, silhouette, dress design, colors, thread pattern, proportions and final composition Duration: 10 seconds Aspect Ratio: 9:16 Style: Ultra-realistic satisfying DIY string-art craft video, cinematic macro photography, realistic artisan hands, 4K. 00:00–00:02 — NAIL OUTLINE Camera: Dynamic macro close-up with a slow top-down push-in. Scene: Clean wooden canvas/frame with the woman's outline marked. Action: The artisan's realistic hands rapidly hammer small nails precisely along the woman's entire outline. Each nail is individually hammered into the canvas with realistic impact and spacing. Show satisfying close-ups of the hammer striking the nails. 00:02–00:04 — BLUE THREAD Camera: Smooth macro tracking movement. Scene: Upper body/bodice area. Action:The artisan begins weaving blue thread between the nails, repeatedly looping and tightening the thread to create the bodice according to the uploaded reference. The thread remains physically connected from nail to nail. 00:04–00:06 — DRESS SKIRT Camera: Smooth cinematic pan downward following the artisan's hands. Action: The artisan rapidly weaves white and gold threads through the lower nails, gradually forming the dress skirt. Show realistic thread tension, overlapping strands and detailed string patterns. The artwork must progressively resemble the uploaded reference. 00:06–00:08 — GOLD DETAIL Camera: Extreme macro close-up with a subtle camera orbit. Action: The artisan creates the final gold circular decorative detail using tightly woven gold thread. The artisan carefully completes the remaining thread connections and trims the excess thread with small scissors. The finished string-art design is now fully formed.00:08–00:10 — FINAL REVEAL Camera: Smooth cinematic zoom-out from macro to full artwork. Action: The artisan's hands leave the frame. Reveal the completed framed string-art artwork. The final artwork must closely match the uploaded reference image exactly — same woman, same silhouette, same dress, same colors, same thread arrangement and same composition. Hold the final artwork clearly for the last second. 🔒 IMPORTANT RULES Uploaded image = exact final artwork reference. The final artwork must be built step-by-step, not appear instantly. Every thread must physically connect between nails. Nails must remain visible where appropriate. Realistic thread tension and weaving. Realistic hammer impacts. Same canvas throughout. No magical appearance. No jump cuts. No morphing No pre-existing finished artwork. Audio: Realistic hammer clicks, thread pulling sounds, scissors snip and subtle workshop ambience. No music, no narration. Negative Prompt: No instant finished artwork, no painting, no brush, no printed image, no digital drawing, no magical appearance, no floating threads, no disconnected threads, no missing nails, no extra hands, no extra fingers, no distorted artwork, no changed design, no color change, no morphing, no jump cuts, no flickering, no CGI, no cartoon, no watermark, no text.
MovieBox-com.community.oneroom-4.0.02.0831.03-90020126.apk
73.1 MB
Ye lo All premium Unlock 🔓 🔓 hai Jav
This media is not supported in your browser
VIEW IN TELEGRAM
Aisa video banana hai to reaction do
Prompt:-
Create an ultra-realistic REAL-WORLD CAMERA VIDEO, not an AI-looking or CGI-looking video. Use TWO uploaded references: REFERENCE VIDEO: This controls the exact car, car movement, camera movement, camera angle, framing, timing, scene transition, environment, lighting, actions and overall visual composition. CHARACTER IMAGE: This controls the exact identity and appearance of the person. The final result must look like an actual real camera recording of a real person inside a real physical Mercedes-Benz car. IMPORTANT – DO NOT GENERATE A 3D RENDER: The car must NOT look computer-generated, game-engine rendered, 3D modeled or digitally reconstructed. The person must NOT look AI-generated, plastic, overly smooth, beautified or digitally composited. Everything must have the small imperfections and optical characteristics of genuine real-world camera footage. REAL CAMERA LOOK: Use physically believable camera optics and real-world photography characteristics. Natural exposure. Natural dynamic range. Natural highlight roll-off. Realistic motion blur. Realistic shutter motion. Natural depth of field. Natural lens distortion. Natural focus transition. Realistic sensor noise/grain at a
Image 1:
characteristics. Natural exposure. Natural dynamic range. Natural highlight roll-off. Realistic motion blur. Realistic shutter motion. Natural depth of field. Natural lens distortion. Natural focus transition. Realistic sensor noise/grain at a very subtle level. Natural skin texture. Natural micro-contrast. Realistic reflections. Realistic shadows. Realistic ambient light. Do NOT make the image excessively sharp, glossy, HDR, oversaturated or digitally perfect. Avoid the typical AI-generated “perfect image” appearance. REAL HUMAN: The uploaded person must look like a real human being physically present inside the car. Preserve the exact identity from the uploaded image: same face shape, same eyes, same eyebrows, same nose, same lips, same jawline, same skin tone, same hairstyle, same facial hair, same age, same natural facial proportions. Preserve realistic skin pores, tiny skin imperfections, natural facial asymmetry and realistic skin texture. Do not smooth the skin. Do not beautify. Do not reshape the jaw. Do not enlarge the eyes. Do not change the nose. Do not modify the lips. Do not make the face symmetrical. Do not make the person look like an AI model. The person’s face must remain stable throughout the entire video.
Create an ultra-realistic REAL-WORLD CAMERA VIDEO, not an AI-looking or CGI-looking video. Use TWO uploaded references: REFERENCE VIDEO: This controls the exact car, car movement, camera movement, camera angle, framing, timing, scene transition, environment, lighting, actions and overall visual composition. CHARACTER IMAGE: This controls the exact identity and appearance of the person. The final result must look like an actual real camera recording of a real person inside a real physical Mercedes-Benz car. IMPORTANT – DO NOT GENERATE A 3D RENDER: The car must NOT look computer-generated, game-engine rendered, 3D modeled or digitally reconstructed. The person must NOT look AI-generated, plastic, overly smooth, beautified or digitally composited. Everything must have the small imperfections and optical characteristics of genuine real-world camera footage. REAL CAMERA LOOK: Use physically believable camera optics and real-world photography characteristics. Natural exposure. Natural dynamic range. Natural highlight roll-off. Realistic motion blur. Realistic shutter motion. Natural depth of field. Natural lens distortion. Natural focus transition. Realistic sensor noise/grain at a
Image 1:
characteristics. Natural exposure. Natural dynamic range. Natural highlight roll-off. Realistic motion blur. Realistic shutter motion. Natural depth of field. Natural lens distortion. Natural focus transition. Realistic sensor noise/grain at a very subtle level. Natural skin texture. Natural micro-contrast. Realistic reflections. Realistic shadows. Realistic ambient light. Do NOT make the image excessively sharp, glossy, HDR, oversaturated or digitally perfect. Avoid the typical AI-generated “perfect image” appearance. REAL HUMAN: The uploaded person must look like a real human being physically present inside the car. Preserve the exact identity from the uploaded image: same face shape, same eyes, same eyebrows, same nose, same lips, same jawline, same skin tone, same hairstyle, same facial hair, same age, same natural facial proportions. Preserve realistic skin pores, tiny skin imperfections, natural facial asymmetry and realistic skin texture. Do not smooth the skin. Do not beautify. Do not reshape the jaw. Do not enlarge the eyes. Do not change the nose. Do not modify the lips. Do not make the face symmetrical. Do not make the person look like an AI model. The person’s face must remain stable throughout the entire video.
Chatgpt prompt 👇👇
Transform the uploaded image into a divine, realistic depiction of Mata Parvati, while preserving the person’s exact facial identity and features. Dress her in an elegant red-pink saree with golden embroidery, traditional jewelry, maang tikka, and long dark hair. Give her a slightly upset, mildly angry and sulking expression—gentle frown, subtle raised eyebrows, calm eyes, natural lips. Add a soft divine aura, cinematic lighting, realistic skin and fabric details, respectful Indian mythological aesthetic, photorealistic, no text, no watermark. 4:5 ratio
Transform the uploaded image into a divine, realistic depiction of Mata Parvati, while preserving the person’s exact facial identity and features. Dress her in an elegant red-pink saree with golden embroidery, traditional jewelry, maang tikka, and long dark hair. Give her a slightly upset, mildly angry and sulking expression—gentle frown, subtle raised eyebrows, calm eyes, natural lips. Add a soft divine aura, cinematic lighting, realistic skin and fabric details, respectful Indian mythological aesthetic, photorealistic, no text, no watermark. 4:5 ratio
Google flow prompt 👇👇
A highly cinematic, hyper-realistic slow-motion shot. Start with a close-up of a metallic trident striking a solid icy glacier. The camera rapidly tilts up and revolves to reveal [YOUR UPLOADED CHARACTER] dressed in a gorgeous traditional Indian red saree and heavy ornate gold jewelry. The character has a fierce, determined expression, with hair blowing dramatically in the strong freezing wind. The background features towering, majestic snow-capped Himalayan mountains under a dramatic cloudy sky. As the camera pans slightly, reveal seven elderly sages (Saptarishi) in the background wearing simple white robes with long white beards, walking towards the main character with their hands folded in respect and pleading. High contrast lighting, 8k resolution, epic fantasy movie style, dynamic camera movement
A highly cinematic, hyper-realistic slow-motion shot. Start with a close-up of a metallic trident striking a solid icy glacier. The camera rapidly tilts up and revolves to reveal [YOUR UPLOADED CHARACTER] dressed in a gorgeous traditional Indian red saree and heavy ornate gold jewelry. The character has a fierce, determined expression, with hair blowing dramatically in the strong freezing wind. The background features towering, majestic snow-capped Himalayan mountains under a dramatic cloudy sky. As the camera pans slightly, reveal seven elderly sages (Saptarishi) in the background wearing simple white robes with long white beards, walking towards the main character with their hands folded in respect and pleading. High contrast lighting, 8k resolution, epic fantasy movie style, dynamic camera movement
Use the uploaded reference video as the EXACT VIDEO REFERENCE. I will upload THREE separate reference images, one for each person: Reference Image 1 = PERSON 1, Reference Image 2 = PERSON 2, Reference Image 3 = PERSON 3. Create a new video that follows the uploaded reference video as closely as possible. IMPORTANT — PRESERVE THE THREE PEOPLE: Use the three uploaded reference images as the identity source for the corresponding persons. Do NOT change, replace, merge, beautify, redesign, or reinterpret their faces. Keep each person's exact facial identity, face shape, hairstyle, hairline, skin tone, body proportions and overall appearance consistent with their reference image. Each person must remain clearly recognizable as the same person from their uploaded reference image. REFERENCE VIDEO MATCH: Recreate the reference video's exact sequence of shots, timing and pacing, camera movement, camera distance, camera height, camera angle, framing, composition, subject positioning, walking and movement direction, hand gestures, head movements, body movements, facial expressions, interactions between people, transitions between shots and overall visual rhythm. Do NOT create a different interpretation of the scene. Do NOT add new actions. Do NOT remove actions. Do NOT rearrange the sequence. Follow the reference video's visual choreography and timing as closely as possible. PERSON CONSISTENCY: When Person 1 appears, use ONLY Reference Image 1 as the identity reference. When Person 2 appears, use ONLY Reference Image 2 as the identity reference. When Person 3 appears, use ONLY Reference Image 3 as the identity reference. Do not accidentally swap identities between shots. Do not mix facial features between the three people. Do not generate generic replacement faces. CAMERA: Match the reference video's handheld smartphone-camera feeling, perspective, camera height, lens look, framing, movement, push-ins, pull-backs, close-ups and wider shots as closely as possible. Preserve the same vertical 9:16 composition. MOTION: Reproduce the same natural human movements shown in the reference video. Keep movements realistic and physically consistent. No exaggerated AI movements. No robotic motion. No unnatural body deformation. Hands and fingers must remain anatomically correct. VISUAL STYLE: Photorealistic live-action footage. Natural human skin texture. Realistic lighting and shadows. Realistic outdoor/environmental lighting. Natural motion blur. Natural depth of field. Real smartphone-camera image quality. No CGI. No 3D-rendered appearance. No cartoon look. No plastic AI skin. No artificial beauty filter. ENVIRONMENT: Preserve and recreate the same environment, background arrangement, atmosphere, lighting conditions, perspective and overall visual mood visible in the reference video as closely as possible. Do not unnecessarily redesign or replace the environment. IDENTITY LOCK: The reference images control WHO the people are. The reference video controls HOW they move, where they appear, the camera behavior, shot sequence, interactions and timing. Do not change the identity of any person just because the pose, camera angle, lighting or expression changes. FINAL OUTPUT: Generate one continuous vertical 9:16 photorealistic live-action video that closely reproduces the uploaded reference video with the three uploaded people replacing the corresponding people in the reference. The result should look like the same video was recreated with these three exact people, not like a new AI-generated interpretation. Maintain strict identity consistency throughout the entire video and preserve the reference video's choreography, timing, camera behavior, composition, movements and interactions as accurately as possible. NO TEXT. NO WATERMARK. NO LOGO. NO CAPTIONS.