Single continuous generation without segment stitching.
Mô hình video AI sắp ra mắt
Seedance 2.5 sắp ra mắt
Seedance 2.5 được preview như một mô hình video đa phương thức cấp sản xuất với clip gốc 30 giây, tối đa 50 tham chiếu, xuất 4K và chỉnh sửa cục bộ thông minh.
Why this page matters
- Native 30-second continuous generation is previewed as the headline upgrade over Seedance 2.0.
- One task can combine up to 50 references across images, videos, audio, 3D blocking, and storyboards.
- The previewed output stack includes 720p, 1080p, native 4K, 10-bit color, Rec.709, and up to 60fps.
- Local intelligent editing is planned for replacing products, people, backgrounds, props, and expressions without regenerating the full video.
- Current platform labels remain Seedance 2.0 based: seedance-fast maps to Seedance 2.0 fast, and seedance-high maps to Seedance 2.0.
Popular public pages
Use the links below to navigate the main public sections of Seedance 3.0 without relying on JavaScript rendering.
HomepageStart generating AI videos from text prompts or image references.MembershipCompare Stars, plans, and creator access options.World Cup 2026Create football-themed photo-op and stadium selfie videos from your photo.CommunityBrowse published AI videos, prompts, and remake ideas.Help CenterRead tutorials, FAQs, and workflow guidance.API DocsConnect bots, scripts, and agent workflows with the public API.Commercial GuideRead commercial AI video guides with copy-ready prompts and workflows.HappyHorse 1.1Explore the supported happyhorse1.1 AI video model and its creation modes.Seedance 2.5Preview the upcoming Seedance 2.5 model with 30s native video and 50 references.Privacy PolicyUnderstand how Seedance handles accounts, data, and protection.Terms of ServiceReview platform rules, acceptable use, and service conditions.
About this page
Seedance 2.5 is not yet available as a selectable model on Seedance 3.0. This page is a public preview so users and crawlers can understand the upcoming capability before the official integration is ready.
The current generator still uses the connected models such as seedance-fast, seedance-high, HappyHorse, and Veo. Seedance 2.5 specs should be treated as upcoming product direction, not current task limits.
The final promotional artwork and demo video are still in production, so the interactive page uses existing site visuals as placeholders until the finished assets are ready.
Key preview metrics
Images, videos, audio, 3D blocking, and storyboard sketches.
Up to 3840 x 2160 Ultra HD, not post-upscaled.
More tonal headroom for commercial color grading.
For sports, product motion, dance, and fast camera moves.
Long script prompts for story, camera, lighting, and constraints.
Detailed capabilities
Native 30-second video
Generate a complete continuous 30-second clip in one inference path instead of stitching short segments together.
- 5s / 10s / 15s / 20s / 30s duration choices
- Continuation from a 30-second result
- More stable characters, lighting, and camera continuity
Commercial image quality
Designed for 720p, 1080p, and native 4K outputs across common social, ad, and cinematic aspect ratios.
- 9:16, 16:9, 1:1, 4:3, and 2.39:1
- Rec.709 workflow compatibility
- Hair, fabric, metal, and small product detail retention
50-reference multimodal input
A single task can combine many kinds of creative references for identity, camera, rhythm, and style control.
- Up to 30 images
- Up to 10 reference videos
- Up to 10 audio references
Cinematic camera control
Professional motion presets and manual camera controls are planned for shots that need more precise direction.
- 50 camera templates
- Speed, rotation, zoom, focus, and depth controls
- Handheld, stabilizer, and dolly-like motion styles
Local intelligent editing
Edit a selected region or object without regenerating the entire 30-second video.
- Replace people, products, backgrounds, props, or expressions
- Keep global lighting and camera motion consistent
- Batch ad variants from one base video
Audio-video sync
Seedance 2.5 is positioned for complete video with synchronized audio, including ambience, dialogue, and music.
- Generated environment sound and background music
- External audio alignment
- Negative prompt field up to 1000 characters
Production parameters
These specs are previewed for the upcoming model and will be reflected in the product only after Seedance 2.5 is officially connected.
| Resolution | 720p, 1080p, and native 4K Ultra HD output. |
|---|---|
| Aspect ratio | 9:16, 16:9, 1:1, 4:3, and 2.39:1 cinematic widescreen. |
| Color | 10-bit color depth with Rec.709 compatibility for editing workflows. |
| Frame rate | 24fps, 30fps, and 60fps high-motion mode. |
| Inputs | Text, image, reference video, audio, 3D blocking, and storyboard sketches. |
Seedance 2.0 vs Seedance 2.5
Current platform labels map to Seedance 2.0 generation: seedance-fast is Seedance 2.0 fast, and seedance-high is Seedance 2.0. Seedance 2.5 is not yet selectable.
| Parameter | Seedance 2.0 | Seedance 2.5 |
|---|---|---|
| Native single clip duration | 15 seconds | 30 seconds |
| Maximum references | 12 assets | 50 full-modal assets |
| Local editing | Not supported | Targeted local redraw across the frame |
| 3D blocking input | Limited support | Full motion path constraints |
| Camera templates | 12 presets | 50 professional templates |
| Batch variants | Limited | One-click multi-version ad variants |
| Long continuation | Stitching can drift | Native coherent continuation |
Commercial use cases
Seedance 2.5 is positioned for teams that need longer, more controllable, and more reusable AI video outputs.
Advertising
Create 30-second brand films, product launches, and multi-version campaign assets.
Short drama
Maintain character, lighting, and camera continuity across narrative clips.
E-commerce
Use product references, batch variants, and local replacement for SKU videos.
Industrial simulation
Use 3D blocking and motion constraints to generate visual simulation data.
Current platform status
This is a preview page only. Seedance 2.5 is not exposed as a model option, and generation still uses the currently connected models until the official integration is ready.
- Use the current generator for seedance-fast, seedance-high, HappyHorse, and Veo workflows.
- Treat Seedance 2.5 specs as upcoming capabilities, not current task limits.
- Replace the placeholder visuals after final promotional images and videos are ready.
Seedance 2.5 capability guide
This guide summarizes the previewed capability groups so search engines, AI agents, and lightweight browsers can understand the planned model positioning.
Longer native video
Seedance 2.5 is previewed with a 30-second native continuous generation limit, doubling the 15-second single-clip limit associated with Seedance 2.0.
More reference control
The model is described as supporting up to 50 full-modal references, including 30 images, 10 reference videos, and 10 audio references in one task.
Production image quality
Previewed output parameters include 720p, 1080p, native 4K, 9:16, 16:9, 1:1, 4:3, 2.39:1, 10-bit color, Rec.709, 24fps, 30fps, and 60fps.
Cinematic editing workflow
Seedance 2.5 is positioned for camera templates, 3D blocking, storyboard input, audio-video sync, and local redraw workflows for commercial variants.
Seedance 2.5 preview FAQ
Is Seedance 2.5 available on this platform now?
No. The page is a preview only. Seedance 2.5 is still described as enterprise beta and is not currently selectable in the generator.
What are seedance-fast and seedance-high today?
The current site mapping remains Seedance 2.0 based: seedance-fast is Seedance 2.0 fast, and seedance-high is Seedance 2.0.
What is the biggest Seedance 2.5 upgrade?
The headline change is native 30-second continuous video generation, plus up to 50 full-modal references and local intelligent editing.
Why are placeholder visuals used?
The final Seedance 2.5 promotional images and demo videos are still being produced, so the page uses existing site visuals temporarily.