Roll for 30 seconds straight
Footsteps land, engines rumble, the score swells on cue. Audio is generated in the same pass as the pixels — matched to on-screen action, never glued on after.

Tell the whole story in one pass — attach faces, products, clips or audio, name them in the prompt, sound included.
Footsteps land, engines rumble, the score swells on cue. Audio is generated in the same pass as the pixels — matched to on-screen action, never glued on after.
Drop in faces, products, wardrobe, locations and style frames — up to 50 multimodal inputs. Direction set in a few frames, applied to all thirty seconds.
Wrong label on the bottle? Background off-brand? Describe the fix and Seedance 2.5 repaints just that region — the rest of the clip stays untouched.
Cinematic motion, fluid camera work and synced audio in one pass. Prompt any aspect ratio from vertical 9:16 to widescreen 21:9.
Up to 30 seconds in one continuous pass — no stitching required, with audio generated alongside the video.
Yes. Footsteps, engines, ambience and score are produced in the same pass as the pixels and matched to on-screen action, rather than added afterwards.
Up to 50 multimodal inputs — faces, products, wardrobe, locations and style frames — and the direction they set applies across the full clip. That is 30 images, 10 clips and 10 audio takes; each clip and each audio take runs 2–30 seconds, and each kind totals at most 30 seconds.
Yes. Describe the change and Seedance 2.5 repaints just that region — swap a label, clean a background or change wardrobe while the rest of the clip stays untouched. On the Edit input the clip is named @Video1 and keeps its own length and shape.
Anything from vertical 9:16 through 16:9 to widescreen 21:9 — plus 1:1, 4:3 and 3:4, or Auto to let the model pick. Animating a still and editing a clip both take their shape from the source.
No. Seedance 2.5 tops out at 1080p; 480p and 720p are the other two options.