Why Seedance 2.0?
Multimodal refs
Up to 9 images, 3 clips, and 3 audio files in one cook.
First–last frame
One still is the first frame. Two stills pin start and end.
Native audio
Dialogue, effects, and music generated with the picture.
Edit & extend
Change an object or continue a scene from a reference clip.
Max 15 sec
Clips from 4 to 15 seconds at 480p, 720p, or 1080p.
Pricing
Billed per generated second. Attaching a video input uses the higher with-video rate for the whole clip.
| Res | No video input | With video input |
|---|---|---|
| 480p | ✦ 12/s | ✦ 15/s |
| 720p | ✦ 25/s | ✦ 30/s |
| 1080p | ✦ 55/s | ✦ 75/s |
How Seedance 2.0 works
Generate high-quality video from text, images, video clips, and audio — all in one pass with synchronized sound. Seedance 2.0 is ByteDance’s next-generation video model, built on a unified multimodal architecture that accepts mixed inputs and produces coherent, audio-synced output.
Outputs are .mp4, at 480p, 720p, or 1080p, 4 to 15 seconds. Set aspect ratio to adaptive, 21:9, 16:9, 4:3, 1:1, 3:4, or 9:16. Native audio is on by default.
What’s new in 2.0
- Multimodal reference inputs — combine up to 9 images, 3 video clips, and 3 audio files. Reference them as [Image1], [Video1], [Audio1].
- Better motion and physics — more realistic sports, dancing, and object collisions.
- Video editing and extension — modify an existing clip or describe what happens next.
- Adaptive aspect ratio — set ratio to adaptive and the model chooses the best fit from your inputs.
Capabilities
Same cook bar for every mode. Picture and sound are generated together.
- text-to-video — skip the trays. The prompt becomes the whole shot, including sound.
- image-to-video — one still is the first frame. A second still is the last frame.
- multimodal reference — 3–9 stills, plus optional video and audio. Tag [Image1], [Video1], [Audio1] in the prompt.
- video editing — attach a clip and describe the change. Motion and camera stay; the edit lands on top.
- video extension — attach a clip and describe what happens next. Characters, set, and style carry forward.
- native audio — dialogue, sound effects, and music, generated with the clip. Turn it off for silent video.
How to use
Everything happens in the cook bar at the bottom of the page.
- 1. Prompt — describe the scene, action, camera, lighting, and mood. Put spoken lines in double quotes. Tag refs as [Image1], [Video1], [Audio1].
- 2. References — skip files for text-to-video. One still is the first frame. Two stills are first and last. Three or more stills are character/style refs. Video refs enable edit, extend, and motion transfer. Audio refs need at least one image or video.
- 3. Settings — ratio, 480p / 720p / 1080p, duration 4–15 seconds. Audio is on unless you turn it off.
- 4. Cook — the button shows the price (resolution × duration; a video input uses the higher with-video rate). Closing the tab after cook starts is fine; we email you when the MP4 is ready.
Text to video
Describe a scene in natural language and get a video with matching audio. The model understands multi-subject interactions, camera movements, and emotional tone. For dialogue, put speech in double quotes — it generates matching lip movements and voice.
Image to video
Animate a still by attaching one image as the first frame. Add a second still to control the last frame. The model preserves the look of your input while adding natural motion. First/last frame cannot be combined with reference-image sets.
Multimodal reference
Combine images, videos, and audio as references. For example, a reference video for motion, stills for character appearance, and audio for rhythm — then describe how to combine them. Tag them in the prompt: “The character from [Image1] performs the dance from [Video1].”
Video editing
Provide a reference video and describe changes — replace an object, change a background, or alter the style. The model preserves the original motion and camera work. Example: “Replace the perfume in [Video1] with the face cream from [Image1], keeping all original motion.”
Video extension
Provide a reference video and describe what should happen next. The model continues the scene with consistent characters, environment, and style.
Tips
- Be specific — camera, lighting, mood, and actions.
- For dialogue:
The man stopped and said: "Remember this moment." - Start with 5-second clips while experimenting, then increase duration.
Settings
-
prompt - Required. Up to 4000 characters. Tag refs as [Image1], [Video1], [Audio1]. Put dialogue in double quotes.
-
images - None for text-to-video. One still is the first frame. Two stills are first and last frame. 3–9 stills are character/style refs. Cannot mix first/last with reference images.
-
videos - Up to 3 clips, 15 seconds combined. Motion transfer, edit, or extend. Tag [Video1].
-
audio refs - Up to 3 files, 15 seconds combined. Needs at least one image or video. Tag [Audio1].
-
aspect_ratio - adaptive, 21:9, 16:9, 4:3, 1:1, 3:4, or 9:16. adaptive lets the model pick from your inputs.
-
resolution - 480p, 720p, or 1080p.
-
duration - 4 to 15 seconds.
-
generate_audio - on by default. Turn it off for a silent clip.
FAQ
What can Seedance 2.0 generate on Soup.video?
Text-to-video, first-frame image-to-video, first–last-frame, mixed image/video/audio references, video editing, and video extension. Native audio is on by default. Combined video refs max 15 seconds, up to 3 clips. Up to 9 stills and 3 audio files.
How do I use Seedance 2.0?
Write a prompt. Attach stills, clips, and audio or skip them. Set ratio, 480p / 720p / 1080p, duration 4–15 seconds, and audio. Cook. Closing the tab after cook starts is fine; Soup.video emails you when the clip is ready.
How does image-to-video work?
Attach one still. That image is the first frame. You can add a second still as the last frame. First/last frame cannot be combined with 3+ reference images.
When should I attach a video reference?
Use video refs for motion transfer, editing, or extending a clip. Tag them as [Video1] in the prompt. Combined length can be up to 15 seconds. With a video ref, Soup uses the with-video per-second rate.
Does Seedance 2.0 generate audio?
Yes. Audio and video are generated together — dialogue in double quotes, sound effects, and music. You can turn it off. Audio refs need at least one image or video and are tagged [Audio1].
What resolutions, durations, and prices does Soup.video support?
MP4 at 480p, 720p, or 1080p, from 4 to 15 seconds. Pricing is per output second. No video input: ✦ 12/s at 480p, ✦ 25/s at 720p, ✦ 55/s at 1080p. With video input: ✦ 15/s, ✦ 30/s, ✦ 75/s.