AI video remake
Upload a reference, analyze its shots, edit the creative direction, and select a model to generate a new video.
Video remake analyzes a reference into shots and uses the edited shot directions to generate a new video. Use it to carry forward camera language, pacing, and scene structure while changing the subject or creative direction.
Input and output
Input is limited to 100 MB and 60 seconds. Use a browser-playable file such as MP4. Analysis produces shots, reference frames, and prompts; generation creates new footage.
Steps
- Upload a reference or choose an available example.
- Select a language and start analysis.
- Inspect shots and edit subjects, scene descriptions, and prompts.
- Choose an available model, quality, and target duration. Check estimated credits and any segment plan.
- Generate, follow segment progress, and preview or download the finished result.
Duration and models
The page offers duration choices based on current model capabilities. Beyond a model's single-segment limit, shots are grouped for linked generation; the plan is shown before generation. Free accounts currently use MiniMax H3 Eco for targets up to 15 seconds, subject to the quota shown on the page.
Analysis and video generation are separate stages. To analyze and export shots only, use Video to Prompt. To download separate pieces of the original footage, use video clipping.
Common questions
Does the previous remake URL work?
The old /video-replicate route redirects to /video-remake.
Why is a longer video divided into segments?
Its target duration exceeds the chosen model's single-segment capability. Shorten the duration or continue with the displayed plan.
How do I replace a product or person?
After analysis, edit the subject information in shot descriptions and prompts while retaining desired camera movements, lighting, and rhythm, then generate.