Source-video workflow
AI Video Editing
A useful edit brief separates the transformation from the invariants. Describe the new visual treatment, then name the timing, subject or composition that should remain recognizable.
Create with Video edit
Open Workspace with this model or workflow selected. Adjust settings, add your media, and create in one place.
Start CreatingChoose between transformation and directed editing
Wan 2.6 video-to-video accepts one to three source videos and produces a fixed 5- or 10-second result. Wan 2.7 video edit accepts one source video plus an optional image and can use a selected 2–10-second length or the full source duration.
Write a bounded edit instruction
Name the target change first, such as environment, material or visual treatment. Follow it with the features that should remain stable. Avoid asking one pass to replace every element in the scene.
- Change: the one transformation the result must make.
- Preserve: subject identity, timing, camera path or composition.
- Avoid: details that must not appear in the edited result.
- Review: whether the requested change fits the source clip's duration and framing.
Choose a source clip that can survive the edit
Use a clip with a readable subject, stable enough compression and no accidental cuts inside the action you need to preserve. If the source timing is already wrong, trim it before generation; an edit prompt is better at changing appearance or direction than repairing an unclear source sequence.
Full-source edits use measured duration
A Wan 2.7 full-source edit is priced from the rounded-up source-video length. The studio shows the total before generation starts.