A reference-video alternative to Higgsfield Genjutsu
Higgsfield Genjutsu presents motion transfer and object swapping as ways to recast an existing video. Genjutsu AI is an independent website for a related creative workflow: reference-guided video generation with MiniMax H3 Max Reference, served through fal. We are not affiliated with Higgsfield and do not run its proprietary Genjutsu model.
How to use this workflow
- Use Genjutsu AI when you want to explore a new character, visual style, or setting from a short reference video.
- Upload one reference clip and optionally one character or style image. Write explicit instructions about each input's role.
- Treat the result as a new generated interpretation. Evaluate movement, continuity, appearance, and sound before publishing.
- Choose a tool with a documented video-editing workflow if your job requires exact object replacement, unchanged backgrounds, or precise source timing.
A prompt to start with
Video 1 supplies the gesture and steady camera. Image 1 supplies the new subject appearance. Recreate the gesture in a softly lit studio, preserving the overall movement and framing. Sound: natural room tone.
What to expect
Similar creative intent does not imply identical capabilities. Genjutsu AI currently offers reference-guided generation; pixel-preserving object swap, batch generation, and exact motion lock are not available.
For the official Higgsfield product, see Higgsfield Genjutsu. Our model's inputs are documented in fal's H3 Max Reference API.
Open the studio ↗Updated October 6, 2026 · Genjutsu AI