Where MiniMax Recast Sits in the Fuser Video Toolset
Recasting lives on this page; for reframing, stylized motion or text-driven scenes, look to MiniMax H3 Reframe, Hailuo 2.3, Kling 4.0 and the other models in the same workspace.
AI Video Prompt Generator

Feedback

minimax recast

Swap the people in an edited clip with MiniMax Recast. Your timing, camera moves, cuts and soundtrack stay intact — just add up to four photos.

All Tools

Discover our comprehensive AI-powered animation toolkit

Recasting Without Rebuilding the Whole Edit

The shot list, pacing and mix stay put. MiniMax Recast handles a single job — deciding who stands in frame — instead of generating the scene all over again.

  • It Edits, It Doesn't Regenerate
    Your existing file is the starting point. Frames, motion and audio are read straight from it, so the change is limited to the people who appear on screen.
  • Acting, Lens and Mix Come Along
    Expressions, head turns, beat timing, lighting, camera movement, cut rhythm and the original soundtrack all carry through the swap untouched.
  • Casting Decided by Upload Order
    Photos are matched by slot, starting at the left of frame, and each identity is tracked across cuts. A short instruction can reassign any of them.

Running MiniMax Recast in Three Steps

Upload the cut, attach portraits, then render — the whole recast is done in three moves.

Where MiniMax Recast Performs Best

Sharp, well-lit, controlled shots are where it shines: the cut and the mix are preserved, and only the on-camera talent is replaced.

Four People, Matched by Position

Four photos is the ceiling, and each one claims a main subject based on screen position unless a prompt overrides it; identities hold from shot to shot.

Same Acting, Different Faces

Head turns, timing, expressions, lighting and lens work all come from your footage; face, hair and wardrobe come from the photos you supply.

Edit Rhythm and Audio Untouched

The pacing of the cut and its original track survive the swap, so the result drops straight into the timeline instead of being rebuilt from zero.

Ask for Faces Only

By default a photo's hair and clothing often tag along, even from a tight headshot; request a face-only swap when the source styling has to stay.

Cheap 768p Check Before 1080p

Run a quick low-res pass to confirm casting and likeness, then finish in 1080p. Runtime equals the source, so credits scale with length.

Best on Clean, Static Subjects

Works on sharp, front-facing, evenly lit people; it cannot invent new motion or sets, and likeness weakens in long shots or profile turns.

FAQ

MiniMax Recast: Common Questions Answered

The answers people look for before swapping the cast of an already-edited video.

1

What exactly does this tool change in my video?

It replaces the on-screen people with the ones in your photos. Motion, camera work, edits and audio are read from your file; matching runs left to right unless you say otherwise.

2

What source length is allowed?

Between 5 and 30 seconds, and no single shot may run past 15 seconds — split longer takes first. The output keeps the source runtime.

3

How many people can I swap, and how are they matched?

Up to four, one photo each. Screen position sets the match by default, and a written instruction can reassign any photo to a different person.

4

Which photos give the cleanest result?

A single sharp, front-facing, evenly lit portrait per person. Avoid sunglasses, hats, group pictures and filtered or low-resolution files; ask for a face-only swap if the styling should stay.

5

Is 768p or 1080p the better choice?

Check the casting in 768p first, then commit the final pass at 1080p. Output length equals input length, so credits follow duration.

6

When is this the wrong tool?

When you need new movement, a different camera angle or another location. Crowds, distant faces, fast action and profile turns all weaken the likeness.

Put MiniMax Recast to Work on Your Next Clip

Upload an edited video, add portraits, and let new faces take over the roles while every camera move, cut and audio track stays as you left it.