Skip to main content
The Higgsfield Genjutsu Motion Transfer node recasts a video with new characters, locations or styles taken from reference images while keeping the motion, camera movement, timing and audio of the source clip. Connect the clip to restyle and one or more reference images that describe the new look, then describe the result in the prompt. The request is sent to the Higgsfield API and the generated video is returned.

Inputs

The source video is validated before the request is sent: a clip shorter than 4 seconds is rejected, and a clip longer than 30 seconds is trimmed to its first 30 seconds. Its audio is carried over to the result. Reference images are uploaded first, and the total number of images across all connected slots cannot exceed 8, even though the node exposes eight slots. Higgsfield’s content moderation can reject the input or the result, which raises an error instead of returning a video.

Outputs

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): f0c30fb93dd2af8d564335290fdb6ff61c38e36ab62cef9641fd1aa81cc10cd2