ByteDance's Seedance 2.0 reference-to-video builds one shot from your own material: up to 9 images, 3 video clips and 3 audio tracks, with no first frame. Output is 4 to 15 seconds at 480p, 720p, 1080p or 4K, with native audio.
Workbench
These examples were written for this endpoint. Replace the prompt and sample media URLs with your own accessible inputs, then check the current quote and model availability before submitting.
- Output5s
cliff-temple-goddess-glance `@Image1` supplies the woman and her wardrobe: a white draped chiffon gown with a deep V neckline and a high slit, a gold rope belt, gold arm cuffs and leaf earrings. `@Image2` supplies the place: a ruined marble temple on a sea cliff at sunset, with poppies between the stones. The camera tracks behind her as she walks down the aisle of columns toward the sun, the wind lifting the long hem and train behind her; she stops between two columns, looks back over her shoulder with a slow smile, and the camera pushes in to a close shot with the sun flaring past her hair. `aspect_ratio` `3:4` keeps the portrait frame, and `generate_audio` is `true`, so the clip comes with its own sound.
- 720p
- 3:4
- audio
- Output6s
character-into-location One reference is the person, the other is the place: the prompt names which image does which job, so the woman from the candlelit portrait arrives on the sunset rooftop in the same outfit, and the drying laundry becomes the thing she moves through.
- 720p
- 16:9
- audio
- Output4s
same-suit-three-cities Wardrobe continuity is the hard half of this one: the same open jacket has to behave like the same garment on a third street corner under different light. The references carry the model and the suit, leaving the prompt to describe only what the wind does and where he walks. Up to nine stills, three clips and three audio tracks go in together, so the cast travels between shots as data rather than as adjectives.
- 480p
- 3:4
- 4s
- Output10s
camera-language-match The prompt names where each element comes from: camera from the clip, set from the image, timing from the track. "low, drifting right, never cutting" tells the model exactly what to extract. The output holds one figure walking away down the corridor from `reference_image_urls`, low and uncut start to finish. `aspect_ratio` `21:9`, `resolution` `1080p`.
- 1080p
- 21:9
- audio
This example starts at $0.07 per output second. The Playground updates the estimate when you change the settings.