Seedance 2.0
A multimodal video model that accepts text, image, video, and audio references together — strong at real-world physics and rhythmic/dance motion, with dialogue, ambient sound, and effects generated frame-matched to the picture.
Excels at
- Accepts up to 9 reference images, 3 reference videos, and 3 reference audio clips at once
- Realistic physics and strong rhythmic/dance motion
- Native audio generated frame-by-frame in sync with the visuals
Signal
Use when
- Complex multi-shot cinematic sequences
- Dance and rhythmic-motion content
- Scenes combining several reference assets into one clip
Avoid when
- Simple single-shot clips where the multi-reference setup is overkill
- Budget-first high-volume runs — use Seedance 2.0 Fast or 2.0 Mini
Prompt used for this sample
A paper boat drifting down a rain-slicked city street at dusk, gentle camera follow, cinematic
Also available
Seedance 2.0 Fast — Same reference-driven workflow, lower credit cost, faster turnaround
Seedance 2.0 Mini — Cheaper again than Fast — for the highest-volume, most exploratory runs
Seedance 2.5 — The flagship step up — native 30s clips, a much larger reference cast, and in-place clip editing
Related models
Veo 3.1
True 4K cinematic video with native, synced audio.
Wan 2.7
Plans the shot before generating it — affordable multi-shot video.
Seedance 2.5
Native 30-second clips with in-place editing.