Video model by ByteDance
Seedance 2.5 renders one take up to 30 seconds, with the sound made in the same pass. In Athana Film Studio you cast your own characters, and they keep the same face in every scene.
Free sign-up · Video on every paid plan
One take
4 to 30 seconds
Sound
Made with the picture
Resolution
480p, 720p, 1080p
Frame
9:16, 16:9, 1:1, 4:3
The model
A whole scene with several cuts renders in one generation. Light, faces and costume hold from the first frame to the last, with nothing stitched.
Footsteps, rain, a crowd and spoken lines come with the picture. Write a line in quotes and the character says it.
Tag a character, a prop or a place with @ and the model matches it. Up to 12 references per scene: photos, motion clips and audio.
Write the cuts with lens, angle and timing. Seedance 2.5 plays them in order inside one generation.
Examples
Each film below is one generation in Athana. Its full prompt is on its page, and our guide shows how we wrote them.
One generation · 21 seconds · 6 cuts
The knight drops to one knee behind his shield as the arrows hit it, and opens the pouch: his wife's ring and his son's little wooden horse. A tear runs down in his eyes. He throws the shield away, draws his sword and charges the archers.
One generation · 25 seconds
Fifteen metres in front of them stands one tall tree, and fifteen metres beyond the tree, thirty metres from the knight, @savage bursts out of the fog and sprints straight at them with a machete, screaming. Only he knows that the rope on the tree is a snare.
Templates
Pick a scene, add your photo and render it with your face.
How it works
Start a project. Seedance 2.5 is the first model in the menu and the default.
Make a character once from a photo or a description. Tag it with @ in any scene and it keeps its face.
Describe it in your own words. Mira, the AI film director, turns it into cuts, camera, sound and performance.
Pick the length, frame and resolution. The price shows on the Render button before any credits are spent.
Versions
| Feature | Seedance 2.5 | Seedance 2.0 |
|---|---|---|
| Longest take | 30 seconds | 15 seconds |
| Sound | Made with the picture | Made with the picture |
| Resolution | 480p to 1080p | 480p to 1080p |
| Best for | Long takes, several cuts, dialogue | Short shots for fewer credits |
Film Studio also has Seedance 2.0 Fast, Seedance 2.0 Mini, Kling 4.0, Wan 3.0 Prime and MiniMax H3. Each project picks its own model.
Where to use it
FAQ
Seedance 2.5 is a video model from ByteDance, launched on July 31, 2026. It makes one take up to 30 seconds with sound, from a text prompt plus reference photos, clips and audio. In Athana it is the default model in Film Studio.
Open Film Studio, cast a character and render a take with sound.
Try Seedance 2.5
