Genjutsu guide

Higgsfield recast: replace the cast, keep the performance

A recast changes the people, places or objects in a shot while keeping the take you already filmed. Here is how it works and what you can swap.

See it in action
Two versions of the same dancer facing each other, one lit magenta and one lime green, on a dark wet stage
Illustrative concept image.
FAQ

Frequently asked questions

What is a recast?

A recast replaces the people, places or objects in footage that already exists while keeping the original motion, timing and camera. In AI video it is done from reference images and a short prompt instead of a reshoot.

Is Higgsfield recast free?

No. Recasting is a paid, GPU-based model and plans on this site start at $19 per month. The examples, guides and prompt templates here are free to use.

What is the difference between a recast and motion transfer?

Motion transfer is the technique that carries the motion of your clip onto new references. A recast is the outcome: a shot where the cast, the location or the look has been replaced.

Can I recast a video I did not shoot?

The model will process any clip you upload, but you need the rights to the footage and permission for the likenesses in it. Recasting a real person without consent is not allowed.

Can I recast two people in the same shot?

Yes. Attach a separate reference set for each performer and name both of them in the prompt.

How long does a recast take?

Prep takes 5 to 10 minutes for a trimmed clip and a clean reference set. The render itself is a single generation.

Try Genjutsu now

Recast any motion with your characters, locations and products, or swap specific elements and keep the rest of the shot.

Higgsfield Recast - Recast Any Video with AI