Edit an uploaded video with a text prompt
Use Edit to change the scene, style, lighting, or selected visual details while guiding Seedance to preserve the important subject and motion. The output keeps the source aspect ratio and duration.
Edit an existing video or extend its final moment with Seedance 2.5. Upload one source clip, write a clear instruction, choose 480p or 720p, and review the generated result before downloading it.
Seedance video editing uses one source clip and one written instruction. A focused request gives the model a clear target and makes the result easier to review.
1
Sign in and upload one video you own or are authorized to process. Use MP4 or MOV, 2–30 seconds, up to 200 MB, at 480p or 720p and 24–60 FPS.
2
State what should change, what must remain stable, and any continuity detail that matters. Use Edit for a visual reinterpretation or Extend for the next segment.
3
Submit the task, monitor its status, then watch the full result before downloading. Retry a failed task only after adjusting the instruction or source material.
Seedance 2.5 supports two prompt-led video editing workflows. Both start from your uploaded clip instead of generating a new scene from text alone.
Use Edit to change the scene, style, lighting, or selected visual details while guiding Seedance to preserve the important subject and motion. The output keeps the source aspect ratio and duration.
Use Extend to create what happens after the source clip. Describe the next action, camera behavior, setting, and continuity cues so the new segment connects naturally.
These are the current Seedance 2.5 video editing requirements exposed by the editor. Check the estimate and selected settings before you submit.
Use a source clip when its subject, framing, or movement is more useful than generating a new scene from scratch. Results should be reviewed before publication or client delivery.
Test a different setting, surface treatment, or lighting mood around a product clip while asking the prompt to preserve the product and camera path.
Create alternate visual directions for a short social clip from the same approved source, then select the result that best matches the campaign context.
Describe a setting or art-direction change while retaining the existing action and composition. Avoid assuming every object or written detail will remain identical.
Explore daylight, overcast, evening, or stylized atmosphere directions with a compact prompt that names the desired lighting and what should stay unchanged.
Edit video with AI by translating a creative brief into one specific instruction. This works best when the prompt separates the intended change from protected details.
Use AI video extension to ask for the next moment of an existing shot. Include an action direction and continuity cues for the subject, camera, environment, and pacing.
A useful prompt names the edit, anchors what should stay stable, and adds clear scene details. Use these examples as starting points for your own work.
Prompt formula: requested change + preserved subject or action + camera or lighting continuity + constraints to avoid.
Replace only the background with a clean sunlit kitchen. Keep the bottle shape, label placement, hand movement, and slow push-in camera motion consistent.
Name the specific item and movement that should stay stable instead of using broad requests such as make it better.
Keep the performer, pose, and camera angle. Restyle the scene as a restrained editorial film look with cool window light and subtle grain; do not add titles or logos.
Style can affect fine detail. Explicitly say which identities, text, props, and framing are most important to preserve.
Continue from the last moment: the cyclist rides forward through the same tree-lined street, the camera follows at the same low angle, and the late-afternoon light stays consistent.
For an extension, describe the next action and repeat the few visual cues that make the new segment feel connected to the source.
Extend the shot by one calm beat as the subject turns toward the window, keeping the same room, wardrobe, lens perspective, and gentle handheld movement.
Short, coherent continuations are easier to review than a request that changes place, cast, action, and camera language at the same time.
The official model page provides Seedance 2.5 background. SeeVideo adds the upload, prompt, credit, task tracking, and result workflow described on this page.