VIDEO-TO-VIDEO AI · FREE STARTER CREDITS
Upload a source clip, keep the motion and timing that matter, and direct a new visual style with a focused AI prompt. Start with available free credits.
MiniMax H316:95s2KFree starter credits may be subject to account, generation, and export limits.
Source and generated result
Use the source video as motion evidence, then change the environment, treatment, lighting, or production style. Compare the uploaded clip with the generated result before you create your own.
Source video
Generated output
Use the uploaded clip as the motion reference. Preserve the walking pace, shoulder turn, and handheld camera rhythm. Reimagine the scene as a rain-lit night market with reflective pavement, focused product lighting, subtle crowd ambience, and a clean ending pose.
A 3-step workflow
Choose a short clip with one clear action, readable subject edges, and stable enough framing to guide the new result.
Name the motion, timing, or camera behavior to preserve. Then describe the new scene, style, lighting, and sound.
Create the transformed clip, then check motion, faces, objects, text, and scene continuity before refining the prompt.
Video-to-video capabilities
Use an uploaded clip to guide action timing, pose changes, and camera rhythm.
Direct a new environment, surface treatment, lighting setup, or production look in plain language.
Separate what should remain recognizable from the parts you want MiniMax H3 to reinterpret.
Start from an existing performance or camera move instead of rebuilding the shot from a blank prompt.
Restyle prompt anatomy
Video-to-video is generative, not frame-locked rotoscoping. Ask the model to preserve motion or rhythm instead of requiring every source pixel to remain unchanged.
Prompt formula
Source motion + details to preserve + new visual treatment + scene change + camera emphasis + sound
Example prompt
Use the uploaded clip as the motion reference. Preserve the walking pace, shoulder turn, and handheld camera rhythm. Reimagine the scene as a rain-lit night market with reflective pavement, focused product lighting, subtle crowd ambience, and a clean ending pose.
This is generative video-to-video, not traditional frame-perfect rotoscoping. Ask for motion, rhythm, or camera behavior rather than pixel-level duplication.
A source clip plus a full style change plus a new subject plus a new environment can conflict. Keep the first test focused on one major change.
If the clip is valuable because of choreography, action timing, or camera rhythm, write those constraints before color, wardrobe, or background style.
Creator and marketing workflows
turn one approved camera move into several visual directions for concept review.
retain a product movement or reveal while testing a different environment and lighting treatment.
use choreography and timing as the motion reference for a new visual world.
transform rough footage into a clearer creative reference before full production.
Free credits and pricing
Try the video-to-video workflow with available starter credits. Before generating, the tool shows the current settings and credit requirement. If you need more generations, compare the one-time credit packs below.
Choose by source asset
| Mode | Start with | Use when |
|---|---|---|
| Video to Video | Existing video | Motion or camera rhythm should guide the result |
| Image to Video | Still image | A known frame should become animated |
| Reference to Video | Subject references | Identity consistency matters more than source motion |
| Text to Video | Prompt only | You want blank-slate generation |
Clear answers
You can try the workflow with the free starter credits available on this site. Current generation requirements and any account or export limits are shown before you create a clip.
It is a source-led workflow in which an uploaded video guides motion, timing, camera rhythm, or scene structure while MiniMax H3 generates a new clip from your prompt.
You can use a source clip to guide motion and request a new environment, lighting style, subject treatment, or production look. The result is generative and may differ from the exact source frames.
Use a short clip with one clear action, visible subject edges, stable lighting, and limited competing camera movement.
First state the motion, timing, or camera behavior to preserve. Then describe the new style, environment, lighting, camera emphasis, and sound.
No. Video-to-video focuses on transforming existing footage and carrying over motion. Reference-to-video focuses on recognizable subject identity in a newly generated scene.
Use image-to-video when you want to animate a still or control an opening frame. Use video-to-video when existing motion is the important starting point.
No. This is generative transformation rather than frame-perfect rotoscoping. Review faces, limbs, objects, text, timing, and continuity in every output.
Upload a source clip, describe what should stay and what should change, and try MiniMax H3 video to video with your available free credits.
Upload a Video to Start