30-Second Storytelling
Generate up to 30 seconds of audio-video content in one pass, with setup, development, turning points, and resolution across connected shots.
Audio-Video Joint Generation
Seedance 2.5 AI Video Generator
Seedance 2.5 is ByteDance Seed's next-generation audio-video joint generation model. Generate up to 30 seconds in one pass, then extend, edit, and direct the story with richer references and more precise control.
30-Second GenerationMulti-Round Extension30 Images10 Videos10 Audio ClipsTimestamp Editing
30s Take
Generate a connected audio-video story up to 30 seconds in one pass
50 References
Use up to 30 images, 10 video clips, and 10 audio clips in one request
Extend and Continue
Add continuation rounds while preserving characters, scenes, pacing, and sound
Edit by Time
Target a timestamp, action, camera move, or audio-video detail
Seedance 2.5 Features
Build longer, more directed audio-video stories with references, continuation, and local control.
Multi-Round Extension
Continue an existing output while keeping the main characters, environment, visual language, narrative pacing, and audio continuity aligned.
50 Multimodal References
Use up to 30 images, 10 video clips, and 10 audio clips together to direct subjects, scenes, styles, props, motion, composition, and sound.
Timestamp-Level Editing
Target an action, camera movement, plot beat, or sound cue in a specific time range while preserving surrounding content.
Professional Direction Controls
Use camera perspective editing, performance blocking, motion references, and clay-render or white-model references to communicate structure and choreography.
Green-Screen and Reference Editing
Replace backgrounds, adapt video, or change selected visual elements while preserving the subject, lighting, clothing, hair, and motion.
How to Use Seedance 2.5
Move from a story brief to a coherent, editable take.
1
Write the Story Direction
Describe the subject, ordered actions, camera language, pacing, environment, dialogue, and sound. For a longer take, divide the prompt into time ranges so the model has a clear narrative arc.
2
Map the References
Attach only materials that carry important information: images for character and design, video for motion and camera intent, audio for dialogue or rhythm, and clay renders for blocking and spatial structure.
3
Generate, Extend, and Refine
Review the complete take, then extend the strongest output or make a targeted edit. Use timestamps for local changes and a new generation only when the overall direction changes.
Seedance 2.5 Demos and Reviews
Independent public videos covering professional AI video production, short films, editing, and filmmaking workflows.
Seedance 2.5 Use Cases
Use reference-rich generation where audiovisual continuity matters.
Short Films and Storyboards
Develop a complete scene with a beginning, middle, and end, then extend the strongest take while preserving its audiovisual language.
Brand Ads and Product Stories
Use product, environment, motion, and audio references together; timestamp edits help revise one beat without rebuilding the concept.
Music, Dialogue, and Performance
Combine audio references with character images and performance direction for music videos, dialogue scenes, stage performances, and choreography tests.
Education and Industrial Visualization
Turn a lesson, process, clay render, or assembly plan into a vivid demonstration with controlled camera paths, blocking, lighting, and object relationships.
Seedance 2.5 Specs & Pricing
What you can generate with Seedance 2.5 on PromptGather, based on the current generator settings.
| Developer | ByteDance |
|---|---|
| Generation modes | Text to Video, Image to Video |
| Inputs | Text prompt, Image, Video, Audio |
| Output | Video |
| Resolution | 720p, 480p, 1080p |
| Aspect ratios | Adaptive, 16:9, 9:16, 21:9, 1:1, 4:3, 3:4 |
| Duration | 4–30 seconds |
| Audio | Optional generated audio |
| Prompt length | Up to 30,000 characters |
| Cost | 34–316 credits per second of video |
| Free to try | Requires credits. The cheapest generation costs 136 credits; new accounts get 30 free credits |
Costs are in PromptGather credits. See plans and credit packs.
Related AI Models
Seedance 2.5 FAQ
What is Seedance 2.5?
Seedance 2.5 is ByteDance Seed's next-generation audio-video joint generation model. It supports up to 30-second generation, multimodal references, multi-round extension, and precise video and audio editing for more controlled storytelling.
How long can Seedance 2.5 generate in one pass?
Seedance 2.5 can generate up to 30 seconds of audio-video content in a single pass. It also supports multiple rounds of extension for longer stories; the official model page highlights two additional extension rounds.
How many reference assets can Seedance 2.5 use?
One generation can use up to 30 images, 10 video clips, and 10 audio clips, for up to 50 multimodal reference assets in total. The exact availability can vary by product surface.
Does Seedance 2.5 generate audio with video?
Yes. Seedance 2.5 is an audio-video joint generation model and can produce audio-video clips together. Audio references can also guide dialogue, rhythm, atmosphere, and other sound-related decisions.
What is timestamp-level editing?
Timestamp-level editing lets you specify when a change should happen, such as revising an action, camera movement, plot beat, or audio cue in a defined time range while keeping the surrounding content consistent.
Can Seedance 2.5 edit green-screen footage and camera perspective?
Yes. The model supports green-screen editing, camera perspective editing, and reference-based editing. These workflows can replace backgrounds or adjust camera decisions while preserving the intended subject and action.
What are clay-render and white-model references used for?
They communicate spatial structure, character poses, motion paths, camera angles, and blocking without requiring a fully rendered scene. Seedance 2.5 can combine that structural reference with separate visual, material, lighting, or character references.
Can I use Seedance 2.5 videos commercially?
Commercial usage depends on your PromptGather plan and the applicable Seedance 2.5 model and platform terms. Review the current terms before using generated videos commercially.
Build the whole take, not just the first shot.
Start with a clear direction, bring the references that matter, and use Seedance 2.5 to move from a short brief toward a coherent, editable video.
More AI Video Models
Compare Seedance 2.5 with the other video models on PromptGather. All models share the same credits, so you can switch without another subscription.
