Same scene, same you, five different universes: Sketch, Cinematic, Fantasy, GTA, Claymation. One photo in, five animated clips out. No filming, no code, about $3 in credits. Here is the exact workflow.
A single clear face photo is the whole 'shoot'. Front-facing, decent light, no sunglasses. This is the identity reference every image is built from.
02
02Step 2
Build the base scene
Feed the photo into Nano Banana Pro (image-to-image) and describe your hero scene: "Photorealistic cinematic film still, the EXACT man from the reference photo, standing center frame waist-up in a busy studio, talking straight into the camera mid-sentence, one hand gesturing, 16:9, no text." This one frame becomes your Cinematic strip AND the master for every other style.
03
03Step 3
Restyle the exact frame
Run the base frame back through Nano Banana with a restyle prompt per style (sketch, fantasy 3D, GTA, claymation, anime, LEGO, anything). The magic words: "Keep the EXACT same composition, same man, same pose" — that is what makes the strips line up as one synchronized scene.
04
04Step 4
Animate each style
Send each styled frame to Kling image-to-video (5s is plenty). Same motion prompt for all: talking to camera, gestures, background people moving, camera locked. Then add the style-lock line from the box below so the style does not melt back to photoreal mid-clip.
05
05Step 5
Stack + label
Stack the clips as horizontal strips on a 1080×1920 canvas (CapCut, Premiere, or ffmpeg) and drop a glowing style label on each strip. Crop each strip slightly above center so heads stay fully in frame. Done: one reel, five universes, all talking at once.
The two prompt lines that do the heavy lifting (copy them)
RESTYLE (image): Restyle this exact image as [STYLE]. Keep the EXACT same composition, same man, same pose with hand gesturing, same mouth open talking to camera, same scene behind him. No text, no watermark.
STYLE-LOCK (video): The entire video keeps the [STYLE] look of the start frame in every frame. Camera locked static, natural mouth movement and hand gestures, people moving subtly in the background.
Want this running for you?
Build it step by step inside AI Creative OS, with me in the thread.
