Copy a motion
MiniMax H3 Max Recast
MiniMax H3 Max Recast replaces the main people in a source video (5 to 30 seconds) with people from reference photos, tracking each person across shots while keeping the original motion, camera, cuts and sound. Up to four reference photos, 768P or 1080P. Powered by MiniMax H3.
Opens the video studio with MiniMax H3 Max Recast ready to use. You can look around as a guest. Running it takes the Pro plan or higher.
What MiniMax H3 Max Recast makes, what it needs and what you get
What it makes
Copy a motion
Makes a character move like the person in a reference video.
What you need
- A video with the motion to copy
- Your character (a picture) up to 4 pictures
Optional: what to change.
What you get
Your character performing the same motion
- resolutions up to 1080P
About MiniMax H3 Max Recast
MiniMax H3 Max Recast makes a character move like the person in a reference video. It is made by MiniMax.
To use it, you provide a video with the motion to copy and your character. You get your character performing the same motion. Options include resolutions up to 1080P.
A 5 s result costs about 18,900 credits. Longer or shorter results cost in proportion, about 3,780 credits per second. It is included from the Pro plan and up. You run it in the video studio, in your browser.
Best for
Typical uses for this kind of model.
- Dance and gesture videos
- Animating a character without rigging
- Matching a performance
Price and performance
Price
about$0.378 / second
3,780 credits / second
- Typical run (5 s of output)
- $1.89 · 18,900 credits
Price by option
- resolution 768P$0.378 / second
- resolution 1080P$0.567 / second
Sampled at the model's default settings; other settings can cost more or less.
10,000 credits = $1. Plans include credits at their own rate; failed or refused runs are not charged. Every model's price
Performance (last 30 days)
Not enough runs yet. We show the success rate and time to result once a model has enough runs to be fair to it.
Last 30 days of customer runs. Success rate = successful runs / (successful runs + runs that failed on our side or the model's side). Runs refused by a safety filter, runs with invalid settings or missing input, runs stopped by the customer and runs without enough credits are not counted as failures. Time to result = from submitting to the finished result, including any queue wait; we show the median. Figures are hidden until a model has 20 runs in the window.
How to use MiniMax H3 Max Recast
- 1
Add a video with the motion to copy and your character
Upload your files, or pick them from your library. Nothing else is needed to start.
- 2
Check the settings
Pick the options you want. The cost updates as you choose, and is shown before you generate.
- 3
Generate your video
Your character performing the same motion. Your result appears in the studio, ready to download.
Use MiniMax H3 Max Recast from the API
One request with your API key runs this model. Long jobs answer with a job id to poll. Each run is billed at $0.378 / second (3,780 credits). API: billed separately, from $5: API calls draw on a prepaid balance on the developer portal, not on a studio plan.
curl -X POST https://api.funai.art/generation/v1/generate \
-H "Content-Type: application/json" \
-H "X-API-Key: YOUR_API_KEY" \
-d '{"model_id":"70e126ec-6266-42a3-9e3a-f7736623ab65","workflow":"video_to_video","workflow_variant":"recast","prompt":"A short description of what to generate.","custom_params":{"video_url":"https://example.com/video_url.mp4","reference_image_urls":"Upload one photo per new person...","prompt":"Optional: who becomes whom, or what to keep or change","resolution":"768P","seed":0,"enhance_realism":false},"is_async":true}'Get an API key Full reference for this model Replace YOUR_API_KEY with your key; the example values are placeholders.
Other models for this job
Same job: copy a motion. Compare before you choose.
Pull motion from a reference video and apply it to new subjects or scenes.
SCAIL-2 is an end-to-end character animation model that drives a reference character from a source video without relying on intermediate pose representations like skeleton maps.
Transfer motion from a video to characters in an image using Dreamactor v2. Great performance for non-human and multiple characters
Wan-Animate is a video model that generates high-fidelity character videos by replicating the expressions and movements of characters from reference videos.