01
Motion Transfer
New character. Same performance.
Recast the performer or rebuild the world around the original movement.
- CHANGE
- Character, wardrobe, location, visual world
- PRESERVE
- Action, timing, framing, camera direction
Replace Anything in Video
Keep the take. Replace the reality.
Keep the motion and camera, and replace the character, product, outfit, object, or world in your video.
Start Video Replace01
Reuse a performance you already like
02
Adapt the same take to another product or cast
03
Explore a new location or wardrobe without rebuilding the shot
Replace the character, product, or world. Keep the take.
02
Object Replace
Swap the object. Keep the shot.
Replace a product or prop without rebuilding the creative from zero.
- CHANGE
- The product, prop, or subject you name
- PRESERVE
- Performer, action, framing, scene rhythm
Three steps. No reshoot.
- 01
Upload
Bring the video whose movement and camera you want to keep.
- 02
Replace
Add references for the new character, product, object, or world.
- 03
Generate
Say what changes and what stays, then create the new take.
- 1
- source video
- A 2–15 second clip supplies the take.
- 11
- reference images
- Available after the source video uses one slot in the 12-file context.
- 5–15s
- output duration
- Choose the render length inside the supported range.
- 768P
- output
- The current output resolution.
REVIEW BEFORE PUBLISHING
Generative transformation still needs a review pass.
AI rebuilds the shot rather than cutting out a layer. Check faces, hands, logos, text, product details, and fast occlusions before publishing.
Video Replace questions
What is AI video replace?
AI video replace takes an existing clip and swaps what is in it — a product, an outfit, a prop, a location, or the person — while the original action, timing, framing, and camera movement stay locked. You are not generating a new video from scratch; you are reusing a take you already shot.
Can I change a product in a video without reshooting?
Yes. Add a reference image of the new product and describe the swap. The performance, camera move, and lighting of the original take carry over, so one shoot can cover a whole product line. Check packaging, logos, and proportions before the result goes into an ad.
Can I change an outfit or the location in a video?
Both. Wardrobe, props, and the surrounding environment can each be replaced from reference images, together or one at a time, while the subject keeps performing the original take.
Is this a face-swap tool?
It goes beyond faces. Replace characters, products, clothing, props, and environments from references. If you only need to swap the character in a video, the dedicated character swap tool is the more direct route.
Does it copy motion exactly?
The source guides action, timing, composition, and camera behavior. The result is generative, so review any movement or detail that must match closely.
Can I replace several things in one video?
Yes. Add up to 11 reference images, number them, and map every replacement clearly in the instruction.
What kind of source video works best?
Use a clear 2–15 second clip with readable subjects, limited occlusion, and the action you want to preserve. Short tests make it faster to correct reference mapping before a final render.
Can I use it for product ads?
Yes. You can adapt a take to a different product, presenter, outfit, or setting. Check packaging, logos, product proportions, handling, and spoken or visual claims before using the result in an ad.
How are credits calculated?
H3 Max is priced by output duration in the current workspace. CloneViral shows the credit amount on the generate button before you submit the render.