Continue an existing clip with AI video extension

Video extension requires a compatible source video. RedVideo prepares the clip for the selected route, and the Core workflow captures a continuation frame that becomes visual guidance for the generated segment.

Wan 2.7 also supports video extension and can expose its own workflow-specific inputs. Accepted container, duration, size, and output settings depend on the active model rather than one universal upload rule.

Write the prompt around the action, camera path, environment, and pacing that should follow the source clip. Mention details that need to remain stable, but focus most of the direction on the new segment.

A source with a readable final moment gives the model a clearer handoff. Abrupt cuts, heavy motion blur, occluded subjects, or a rapidly changing camera can make a seamless continuation harder to infer.

• Continue the subject’s visible action
• Maintain or deliberately change the camera movement
• Describe the next event and the intended pace

AI extension is probabilistic and does not guarantee an invisible edit. Compare subject shape, lighting, camera speed, background geometry, and motion at the transition before combining the clips in a final sequence.

If the join drifts, shorten or simplify the prompt, choose a cleaner source ending, or try another supported model. Preserve the original clip so each iteration can start from the same source.

The continuation has model-specific duration, resolution, audio, access, and credit requirements. Repeating the workflow creates another generation rather than an unlimited extension of arbitrary length.

Review the current quote and controls in the studio before each run. If the goal is a different ending rather than more footage, first-and-last-frame guidance may offer a more direct route.