Seedance 2.0
ByteDance's Seedance 2.0 turns text, images, and references into video with native synced audio, up to native 1080p.
Made with Seedance 2.0
Sample outputs. Open in Studio to generate your own.
Where Seedance 2.0 shines
Text to video
Describe a scene in a prompt and generate a video clip with native synchronized audio from 4 to 15 seconds.
Animate a frame
Turn a starting first-frame image into motion, with an optional last frame to control first-last-frame transitions.
Reference composition
Combine up to 9 reference images, 3 videos, and 3 audio clips in Edit mode to direct the look, motion, and sound.
Strengths
- Native synchronized audio generated together with the video, toggleable via a Generate audio switch
- Multimodal references in Edit mode: up to 9 images, 3 videos, and 3 audio clips, addressable with @Image and @Video tokens
- Native 1080p output on the Pro tier, plus 480p and 720p options
- Wide aspect-ratio coverage including 21:9 cinematic, 9:16 vertical, and adaptive
- Flexible clip duration from 4 to 15 seconds
- Text-to-video, image-to-video, and first-last-frame modes in one model
Trade-offs
- Priced per second of output, so longer clips cost proportionally more
- Higher resolutions cost more per second, with 480p cheaper than 720p and 1080p
- Supplying reference videos adds a per-second surcharge
- Image, video, and audio reference inputs are only available in Edit mode, and reference audio requires at least one reference image or video
At a glance
About Seedance 2.0
Seedance 2.0 is ByteDance's multimodal AI video model, available on HexGen as the Pro tier with native 1080p output. Start from a text prompt, animate a single starting frame, or guide the motion between a first and last frame. Every clip can be generated with native synchronized audio baked in, so you get sound and picture together in one pass.
What sets Seedance 2.0 apart is how much you can show it. In Edit mode you can supply up to 9 reference images, 3 reference videos, and 3 reference audio clips, then point the prompt at them with @Image and @Video tokens to compose exactly the look, motion, and sound you want. Outputs span 480p, 720p, and native 1080p, with aspect ratios from cinematic 21:9 to vertical 9:16, plus adaptive.
Clip length is flexible from 4 to 15 seconds, and pricing follows the work done: a per-second rate by resolution, with a small per-second surcharge when you add reference videos. Run it on HexGen with no setup, get MP4 files back, and only pay for the duration and resolution you actually generate.
Starting points
Copy, tweak, and run. Good prompts get you most of the way there.
A neon-lit Tokyo alley at night in the rain, camera slowly tracking forward past glowing signs, ambient city sounds and soft rain, cinematic 21:9, 8 seconds.
Animate this product photo: a coffee cup on a wooden table, steam rising and morning light shifting across the surface, gentle cafe ambience, vertical 9:16.
First frame: a closed wooden door. Last frame: the door fully open to a sunlit garden. Smooth push-in transition with birdsong, 6 seconds, 16:9.
Pay only for what you render. 1 USD = 1,000 HGcoins. HGcoins never expire and failed runs refund automatically.
Seedance 2.0 vs other models
Seedance 2.0 is the Pro tier of ByteDance's Seedance video family, built for native 1080p and rich multimodal references. Here is how it sits next to its faster Turbo sibling and a leading Kuaishou video model.
| Model | Quality | Speed | Cost | Choose it when |
|---|---|---|---|---|
Seedance 2.0 This ByteDance | Best | Fast | Higher cost | Pick this when you want the highest-quality native 1080p output with native audio and the fullest multimodal reference control, up to 9 images, 3 videos, and 3 audio clips. |
ByteDance | Great | Fastest | Mid cost | Choose Turbo for faster, lower-cost runs from the same Seedance family when you do not need the Pro tier's top-end output. |
Kuaishou | Great | Fast | Mid cost | Consider Kling 2.6 if you prefer Kuaishou's video model for an alternative look and motion style. |
Related models
Frequently asked questions
It generates video from a text prompt, animates a starting first-frame image, controls first-last-frame transitions, and composes from multimodal references. It can also generate native synchronized audio together with the video.