Seedance
Long single takes with sound, and references you address in the prompt
Seedance is ByteDance’s video line. Every release generates its soundtrack with the picture, and 2.5 and 2.0 take references as *inputs to the prompt* — you attach images, clips and audio and address them in the text as [Image 1], [Video 2]. 2.5 runs the longest takes in the family; 2.0 adds 4K.
Strengths and limits
Key Features
- 2.5 renders a long continuous shot in one generation, so a walk or a camera move does not need stitching.
- Multimodal references on 2.5 and 2.0: drive one generation from several images, source clips and audio tracks at once.
- The soundtrack is generated with the picture on every release, and can be switched off for a silent clip.
Where it falls down
- Fast and Mini stop at a lower resolution than the standard rows, and 2.5 has no 4K option — for a 4K master use the standard 2.0 row.
- Mini takes a first frame but no reference inputs.
- Multimodal mode requires a prompt that actually references the inputs; passing references without addressing them wastes them.
Model
Capability tables and prices are read from the live model catalogue.
| Model | Tools | Duration | Resolution | Audio | Credits |
|---|---|---|---|---|---|
| Seedance 2.5 | Text-To-Video | 5–30s | 480p · 720p · 1080p | from 13 credits · ≈ $0.74 | |
| Seedance 2.0 | Text-To-Video | 5–15s | 480p · 720p · 1080p · 4k | from 9 credits · ≈ $0.51 | |
| Seedance 2.0 Fast | Text-To-Video | 5–15s | 480p · 720p | from 7 credits · ≈ $0.40 | |
| Seedance 2.5 | Image-To-Video | 5–30s | 480p · 720p · 1080p | from 13 credits · ≈ $0.74 | |
| Seedance 2.0 | Image-To-Video | 5–15s | 480p · 720p · 1080p · 4k | from 9 credits · ≈ $0.51 | |
| Seedance 2.0 Mini | Image-To-Video | 5–15s | 480p · 720p | from 5 credits · ≈ $0.29 | |
| Seedance 2.0 Fast | Image-To-Video | 5–15s | 480p · 720p | from 7 credits · ≈ $0.40 |
Tools
Start from text
Start from an image
Frequently Asked Questions
You attach several images, videos and audio clips to one generation, then refer to them in the prompt by index — "[Image 1] walks into the room lit like [Image 2], scored to [Audio 1]". The model composes from those inputs instead of guessing from words alone.
Ready to Get Started?
Every tool shows its price in credits before you run it.