Seedance
by ByteDance
Multi-shot video from text or an image, in 1080p.
What it is
Seedance is ByteDance Seed's video model, officially "a model that supports multi-shot video generation from both text and image". The multi-shot part is the point: it natively generates narrative video with several cohesive shots while holding the subject, the visual style and the atmosphere steady between them.
That is unusual. Most video models give you one shot and leave the continuity problem to you. Seedance is built to keep a scene recognisable across cuts, which is exactly the hard part of anything longer than five seconds.
ByteDance describes 1080p output with "smooth motion, rich details, and cinematic aesthetics", stable control over multi-agent interactions and complex action, and a stylistic range that runs from photorealism and cyberpunk to illustration and felt texture. It takes both a prompt and an image as a starting point.
What it is good at
Narrative video with multiple cohesive shots, holding the main subject, the style and the atmosphere. The thing that normally takes a script and a lot of re-rolling.
ByteDance calls out "smooth generation of large-scale movements", from a subtle expression to an active scene. Useful when something actually has to happen on screen.
Stable control over multi-agent interactions and complex action sequences , two people, a crowd, a hand and an object.
Photorealism, cyberpunk, illustration and felt texture are all named as supported. A house style is something you can ask for rather than fight.
Where it struggles
- ByteDance publishes no list of limitations for the model, so we do not put words in their mouth. What we can say is measured here: a clip of "four seconds" comes back at 4.06 seconds, and 1080p costs more than twice what 720p does.
- It is priced per thousand tokens on height, width and duration, so resolution drives the bill harder than length does. The Studio works that out and shows the amount before you run.
How to prompt it
Describe the shots, not just the scene. Because it is built for multi-shot, saying what happens first and what happens next gives it something to work with. Keep the subject description identical between shots and change only the action and the camera , that is what holds a sequence together.
An example that shows what that means
A woman in a grey wool coat steps off a tram into a wet street. First shot: from behind, at hip height, as she steps down and the doors close. Second shot: a tighter front angle as she looks up at a building, rain on her shoulders. Overcast light, muted colour, handheld.
What we offer here
Straight from the catalogue, so this list never goes stale. The price is the provider’s published rate with our markup on top, and it is on the button before you run anything.
Questions
What does multi-shot actually mean?
It generates several cohesive shots in one go, holding the subject, style and atmosphere between them. Most models give you one shot and leave continuity to you.
Text to video or image to video?
Both, officially supported. Starting from an image gives you a first frame you have already approved, which is the strongest way to control how a clip opens.
Why is 1080p so much more expensive than 720p?
The provider bills on height times width times duration, so resolution costs more than length. We pass the published rate through with our markup, and the amount is on the button before you run.