1. Add your source video
Paste a public HTTPS URL to the video you want to transform into the video reference field. The clip should be reachable at that URL for the generation to read it.
Use images for identity, video for movement, and audio for timing. The completed generation appears here.
MiniMax H3 Video to Video turns an existing clip into a new generation: add the source video as a reference, describe the change, and the model produces a result guided by that original motion and timing.
The tool above runs on the same MiniMax H3 reference-guided engine used across this workspace. A video reference anchors motion and timing to your source clip, while the prompt directs what should change: style, subject appearance, lighting, or mood. You can add an image reference to control identity or an audio reference to shape timing in the same generation.
Because the source is a whole clip rather than a still frame, video-to-video requests work best when the prompt focuses on the change rather than re-describing what the reference video already shows. State the new treatment clearly and let the video reference carry the motion.
Follow these four steps in the live tool above.
Paste a public HTTPS URL to the video you want to transform into the video reference field. The clip should be reachable at that URL for the generation to read it.
Describe the change, not the source: the new style, subject treatment, or mood you want. State that the video reference should guide motion and timing so the request is unambiguous.
Set duration, resolution, and aspect ratio for the intended output. A shorter 768P pass is a useful first test before committing to a longer or 2K generation.
Submit the task and the tool tracks status until the result is ready. Review the output against the source clip, then adjust the prompt or add an image or audio reference if another pass is needed.
Shift the lighting, color, or overall visual treatment of an existing clip while the video reference keeps the original motion.
Adjust a subject's appearance while a video reference anchors the underlying motion and timing to the source.
Turn a plain clip into something with cinematic lighting, grading, or camera feel guided by the source motion.
Re-render a short clip with a new style or treatment for a specific platform format, using the original as a motion guide.
Both pages use the same generator. The difference is framing, not capability.
| Aspect | MiniMax H3 Video to Video | MiniMax H3 Reference to Video |
|---|---|---|
| Generator | Same live tool | Same live tool |
| Primary framing | Video-first: transform an existing clip | Equal emphasis on image, video, and audio sources |
| Typical use | Restyling, cinematic transformation, character change | Character videos, product videos, multi-source campaigns |
| Input method | Public HTTPS video URL, plus optional image and audio | Uploaded images or public HTTPS image/video/audio URLs |
| Credits | Same duration and resolution pricing | Same duration and resolution pricing |
The video reference already carries motion and timing. Focus the prompt on what should change: style, lighting, subject appearance, or mood.
If you add an image or audio reference alongside the video, state which source controls motion, which controls identity, and which controls timing.
A concise, single-action reference video produces clearer results than a long clip with several unrelated movements.
Confirm the style and motion transfer are working before committing to a longer duration or 2K resolution pass.
Credits are calculated from duration and resolution, and the live generator shows the current estimate before you submit.
Test with a short duration and 768P before scaling to a longer or 2K generation. The same account balance works across every MiniMax H3 workflow.
It is a way to generate a new MiniMax H3 video guided by an existing clip: add the source video as a reference, describe the change, and the model produces a new result guided by that motion and timing.
Paste a public HTTPS URL to the clip in the video reference field on this page. The video must be reachable at that URL when the task runs.
Yes. You can add image and audio references alongside the video in the same generation, and state in the prompt which source controls identity, motion, or timing.
They use the same generation engine. This page is set up for a video-first workflow: transforming an existing clip. The Reference to Video page presents the same tool with equal emphasis on image, video, and audio sources for broader reference-guided generation.
Credits are calculated from the selected duration and resolution, shown as a live estimate before you generate. The same credit balance works across every MiniMax H3 workflow.
That is the goal of using a video reference: the prompt directs the new style or subject treatment while the video reference anchors motion and timing to the source. Results depend on how clearly the prompt separates the two.
Move from research to a relevant generator, prompt resource, or pricing page without restarting your workflow.
Create with image, video, and audio references.
Write and test prompts for specific video goals.
Follow the complete generation workflow.
Understand the speed-focused search intent and availability.
Camera movement, shot composition, and motion control prompting.
Reapply movement from a video reference onto a new subject.
See every online input path in one workflow map.
Understand LoRA availability and the closest online option.
See what a ComfyUI-based path involves and the online alternative.
What API access means here and the online generator alternative.
VRAM, GPU, Mac, AMD, and model size questions in one place.
What running MiniMax H3 locally would involve, mapped out.
Where to look for repositories, workflows, and implementations.
What to look for on Hugging Face: weights, variants, and LoRA.
Open the generator with MiniMax H3 selected.
Compare plans and one-time credit packs.