5 s15 s30 s
Up to thirty seconds
Anything from four to thirty seconds, long enough for a spot to open, turn and land.
Skim low over the waves toward the sea cliffs, white water breaking under the camera, bright afternoon light.Skim low over the waves toward the sea cliffs, white water breaking under the camera, bright afternoon light.
Generate
Seedance 2.5. Thirty seconds, with sound.
ByteDance’s video model, in your workspace. Write the shot, set a start and an end frame or cast it with references, and get up to thirty seconds at up to 1080p, sound included.
Generate with Seedance
Free to start. No credit card.
- DURATION
- 4–30 s
- 5 s by default
- RESOLUTION
- Up to 1080p
- 480p · 720p · 1080p
- ASPECTS
- Auto + 6
- from 21:9 to 9:16
- REFERENCES
- Up to 30
- images, plus 10 clips and 10 tracks
Room for the whole spot.
Long takes, sound and a full cast of references, from one prompt box.
Sound in the same run
Audio is on by default, so the clip comes back with its own sound. Switch it off for a silent take.
A cast, not a sentence
Up to thirty images, ten clips and ten audio tracks steer the people, the product and the place.
A whole city in one clip.
Seedance 2.5 can cut between shots inside a single clip: the bridge from the water, the towers in the fog, a cable car on the hill, the skyline after dark.
This one was made with Seedance 2.5 in the workspace.
Generate with SeedanceWhere it starts. Where it lands.
Drop an image into the Start frame and say what moves. Add an End frame and the clip arrives on it; the prompt says how it gets there.
Start from a frame
Name what goes where.
Three or more images, or any clip or track, turn the run into references. Each one gets a handle in the prompt, @Image1, @Image2, and the sentence puts it in the shot.
Start from references
Seven shapes, Auto included.
Leave the ratio on Auto, or set it before you press: 21:9 for a banner, a square for the feed, 9:16 for Stories.
Generate with SeedanceHow to use Seedance 2.5
Three steps in Video Generate.
Introducing Seedance 2.5
Generate with SeedanceByteDance’s video model for long takes: up to thirty seconds at up to 1080p, sound in the same run, and a cast of up to thirty references.
FAQs
ByteDance’s video model for text-to-video, image-to-video with start and end frames, and reference-to-video. In MarketingX it runs in Video Generate, on your plan.
From 4 to 30 seconds, picked in the Duration chip. 5 seconds is the default.
480p, 720p and 1080p, with 720p as the default. The ratio is Auto by default, or 21:9, 16:9, 4:3, 1:1, 3:4 or 9:16.
One or two images are read as the start and end frames. Three to thirty images, or any clip or audio track, are read as references, with up to ten clips and ten tracks.
Yes. Audio is on by default and the clip comes back with sound; switch it off for a silent take.
One. Press Generate again for another take.