Genjutsu AI Video Generator for image-to-scene studies

Recast both fighters and restage the fight in the pastel valley — every move and camera stays

Replace the two men in the video with the two people from the reference images. Person 1 replaces the man on the left and Person 2 replaces the man on the right. Keep the original camera movement, timing, lighting, and body performance of both men, including their interactions and positions in the frame. Preserve Person 1's bald head, beard, black glasses, white t-shirt, light blue jeans, white sneakers with burgundy detail, and gold chain. Preserve Person 2's face, hair, and outfit exactly as shown in the reference images. Keep both identities distinct and consistent throughout the video, and do not mix or swap their features.

Replace only the woman in the source <<<video_1>>> video with Chloé, using the uploaded Chloé <<<image_1>>> image as the strict identity, face, hair and eyewear reference. This is an identity swap, not a scene regeneration: repaint nothing outside the woman's body, face and hair. Preserve the source performance and shot exactly: the fixed vertical low-angle camera, timing, pacing, crouched opening close to the lens, backward walk into the full-body pose, front pose, side and back turns, bent-leg pose, hair-touch and hair-flip gestures, walk toward the camera, final crouch and hand-under-chin pose. Preserve her body path, pose, gaze direction, facial expressions, hand and leg movements, perspective and scale changes from close-up to full body and back. Keep the original outfit and all scene elements unchanged: the oversized charcoal-gray coat, light cream top, short dark skirt, sheer black tights, black flats, large structured brown handbag, pale stone arcade, pavement, daylight and shadows. The ornate wooden double doors must stay pixel-faithful to the source: their upper two-thirds are glass panes divided by wooden mullions into a grid, with an arched fanlight of radial glazing bars on top; only the lower third is solid carved wood. The glass shows real reflections of the sky and the buildings across the street. Do not fill the glass with wood, do not add or remove panels, do not redesign any molding, mullion or material, do not reinterpret the architecture in any way. The sunglasses come exclusively from the uploaded Chloé image and they are a real product: reproduce them exactly as they appear there - frame shape, lens shape and tint, frame thickness and color, temple arms, and any logo or metal detail. Do not redesign, enlarge, shrink, recolor or restyle the eyewear in any way. The frames must stay perfectly recognizable as the same product in every frame of the video, from close-up to full body.

Transfer the single female performer's exact motion in @Video1 to our original adult AI influencer shown in @Image1 and @Image2. These images depict ONE identical woman: @Image1 is the face identity anchor and @Image2 establishes her appearance and body proportions, not the background or wardrobe. Apply her recognizable facial geometry, natural green eyes, full lips, freckles and long chestnut brown hair consistently throughout the video. Preserve the source video's indoor setting, original outfit (black graphic T-shirt and light-colored bottoms), composition, camera, timing, gestures and dance motion. Match the source illumination and natural motion blur. Keep the original video audio if supported. Output the exact 8-second source segment; no introduction, no additional people or cuts, no studio background. Photorealistic natural skin detail with temporally consistent identity.

Replace the video's main character

@Image1 — WOMAN, the replacement WOMAN: fair skin, brown eyes, bold brows, beauty marks on her cheek and jaw, glossy warm-nude lips, a turquoise-blue blunt bob with straight bangs and subtle aqua-to-lavender variation, small silver earring, and rings; natural young-adult build and stature matched to the original framing. Wardrobe: plain clean regular-fit white cotton T-shirt, plain gray straight relaxed-cut sweatpants, and gray-white sneakers; no sunglasses, prints, logos, or accessories beyond the earring and rings. Video edit. Keep this @Video1 clip exactly as it is — the same shots and cuts (cuts at 0.7s, 2.1s), the same camera moves, framing and composition, the same wood-paneled vintage diner with yellow booths, burger posters, pendant lights, kitchen doors, chihuahua in its sweater, knight patron, masked patron, smartphone, and night drag-race footage, the same warm lighting, pacing and timing of every shot. Change only the dancing man in the black leather vest into WOMAN, keeping original blocking, poses, target motion, positions, screen placement and timing while rendering the replacement person's complete image-defined appearance and stature. 1. REPLACE the original dancing man in the black leather vest (0.7-2.1s over-the-shoulder phone shot; 2.1-9.8s tracking shot through the diner and kitchen doors) with WOMAN from @Image1. Put WOMAN into every one of those shots, matched beat-for-beat to the original holding the smartphone, pocketing it, dancing, lip-syncing, gesturing, moving backward, and pushing through the doors. Render WOMAN's fair skin, brown eyes, bold brows, cheek and jaw beauty marks, glossy warm-nude lips, and consistent turquoise-blue blunt bob with straight bangs and subtle aqua-to-lavender variation; her eyes remain visible with a deadpan confident expression. Keep her neutral white T-shirt, plain gray sweatpants, gray-white sneakers, earring, and rings consistent across all shots; carry the smartphone naturally in her ringed hands. Identity lock: WOMAN maps only to the dancing man in the black leather vest during 0.7-9.8s. Within 0.7-9.8s, replace the original source person identified as the dancing man in the black leather vest in @Video1 completely with WOMAN in every appearance; the original identity must not appear within those windows. Retain the original performance, pose, blocking, interactions, and timing. Outside those windows, preserve that source person's original appearance, subject only to other explicitly requested edits. Lock WOMAN's complete image-defined identity, turquoise-blue bob, visible eyes, neutral wardrobe, and rings across frames and cuts only within 0.7-9.8s; no face or hair-color morphing. WOMAN is never duplicated onto another figure. Protect the smartphone contact and adapt the inherited choreography naturally to WOMAN's stature. Everything else — the night drag-race footage, diner, background patrons, chihuahua, knight, masked patron, smartphone content, warm lighting, color grade, the camera moves and all timing — stays exactly the same.

Use my uploaded photo as the exact identity reference. Replace the person who is signing the car papers with me. I must become that person throughout the entire video. Wherever the original person appears — signing the documents, standing up, walking toward the car, approaching the car, and getting near the car — I must appear instead. IMPORTANT: Keep my exact face, my natural facial features, my exact hairstyle and my own natural hair from the reference photo. Keep my own body and natural body proportions. Do not copy the original person’s body shape. My body should look like my real body from the reference photo. My face must look completely natural and realistic, as if I was actually there during the real filming. Preserve my identity consistently throughout every shot and every camera angle. Do not change my facial structure, eyes, nose, lips, jawline, skin tone or hairstyle. Keep the original video’s: camera movement camera angles framing timing actions gestures walking signing the car documents movement toward the car interaction with the car environment lighting shadows background scene transitions Do not change the story or create new actions. Only replace the original person with me. When the original person signs the car papers, I should naturally sign them instead. When the original person walks toward the car, I should naturally walk toward the car instead. When the original person reaches the car, I should be the person standing next to and interacting with the car. Make the face-to-body connection completely natural. No face-swap artifacts, no distorted facial features, no plastic skin, no artificial appearance, no identity drift. The final result must look like a real-world professional camera recording of me actually buying the car and being present at the dealership. Photorealistic, natural skin texture, realistic facial expressions, realistic hair, realistic body movement, natural lighting and shadows.

Replace the main characters with the characters from my references.