30-Second Storytelling
30-Second Storytelling
Build a Complete Scene
Use a longer generation window to develop setup, movement, camera changes, performance, and resolution inside one coherent sequence.
SEEDANCE 2.5
Seedance 2.5 AI Video Generator
Create longer, more controllable AI videos with Seedance 2.5. Generate complete scenes up to 30 seconds, direct synchronized audio and visuals, use images, video, and audio as creative references, and refine results with stronger editing and production-focused control.
Latest major Seedance modelReleased July 2026
Seedance 2.5 media preview
Seedance 2.5
Verified showcase pending
Create with Seedance 2.5
Turn an idea or reference into video with the Seedance 2.5 generator. Choose the workflow available in the tool, add your prompt and supported reference media, then adjust the available generation settings before creating your result.
Loading your workspace…
See Seedance 2.5 in Action
Seedance 2.5 is designed for more complete video sequences, not only isolated moments. Explore examples that show longer motion, cinematic camera direction, multimodal references, native audio, and controlled scene progression.
Multimodal References
Multimodal References
Direct with Reference Media
Guide characters, products, environments, movement, camera language, style, or sound using supported image, video, and audio references.
Audio + Video
Audio + Video
Generate Sound with the Visuals
Create audio and video together so dialogue, ambience, effects, performance, and visual timing can work as part of the same generation.
Model overview
What Is Seedance 2.5?
Up to 30 secondsMultimodal referencesEditing + extension
Seedance 2.5 is ByteDance Seed’s next-generation audio-video creation model, officially released in July 2026. It builds on the multimodal foundation introduced with Seedance 2.0 and moves the workflow further from generating a short isolated clip toward directing a more complete piece of video.
01
Longer, reference-led creation
The model can generate videos up to 30 seconds in a single pass and supports continued extension for longer sequences. It can also work with a much richer set of reference materials, including images, video clips, and audio.
This makes Seedance 2.5 useful when a prompt alone is not enough to communicate the intended character, product, movement, camera style, voice, environment, or visual direction.
02
Production-focused control
The biggest change is control. Seedance 2.5 combines longer storytelling, multimodal reference understanding, synchronized audio-video generation, targeted editing, and more production-oriented tools in one model.
For creators making ads, narrative scenes, character content, product visuals, or previsualization, that means more of the creative brief can live inside a single generation workflow.
Release highlights
What’s New in Seedance 2.5?
Seedance 2.5 expands several parts of the video-generation process at the same time.
01
Longer generation
First, single-pass generation increases to as much as 30 seconds. A longer window gives the model room to establish a scene, develop an action, change framing, move between story beats, and reach a more complete ending without requiring every idea to be broken into very short clips.
02
Broader references
Second, reference handling is significantly broader. Seedance 2.5 can use up to 30 images, 10 video clips, and 10 audio clips as reference materials in one generation.
These references can help define subjects, scenes, visual style, voices, motion, composition, camera behavior, or other aspects of the intended result.
03
More precise editing
Third, editing becomes more precise. The model supports time-based instructions for changing selected parts of a generated or referenced video while preserving the surrounding sequence.
More advanced workflows include reference-based editing, camera-perspective adjustments, green-screen creation, and production controls designed for more complex visual work.
Together, these changes make Seedance 2.5 less about generating one attractive shot and more about directing a complete audiovisual sequence with clearer constraints.
Seedance 2.5 vs Seedance 2.0
Seedance 2.0 established the unified multimodal workflow that supports text, image, video, and audio input. Seedance 2.5 extends that foundation with longer generation, richer references, and more precise editing.
Single-pass storytelling
Seedance 2.5
Up to 30 seconds with stronger long-form continuity
Seedance 2.0
Designed around shorter multimodal video generation
Reference workflow
Seedance 2.5
Expanded reference capacity and stronger understanding of creative intent, motion, framing, and camera language
Seedance 2.0
Multimodal image, video, audio, and text reference workflows
Editing
Seedance 2.5
More precise time-based editing plus advanced production-oriented controls
Seedance 2.0
Multimodal generation and editing with less emphasis on long-form targeted revision
Best fit
Seedance 2.5
Longer stories, complex references, iterative editing, and production-focused workflows
Seedance 2.0
General-purpose multimodal creation when the extra 2.5 workflow is not required
Choose Seedance 2.5 when longer continuity, complex reference material, targeted revisions, or production control removes meaningful steps from your workflow. For simpler multimodal video tasks, Seedance 2.0 can still be a practical option.
Creative workflows
Best Use Cases for Seedance 2.5
Seedance 2.5 is especially useful when a video needs continuity, references, audio, or a more deliberate sequence of events.
01
Use it for cinematic scenes that need controlled camera movement and multiple connected beats. Use it for product advertising when a reference image must stay recognizable while the camera, lighting, action, and environment evolve around it.
02
Character-focused projects can use visual and audio references to guide appearance, voice, performance, and dialogue.
03
The model also fits short narrative content, fashion films, music-driven visuals, branded social video, storyboard exploration, film previsualization, educational sequences, and creative concepts that benefit from a longer generation window.
04
For very early ideation or large numbers of disposable variations, a faster or lighter model may be more efficient. Seedance 2.5 is most valuable when the additional control directly improves the final result or reduces manual assembly.
Media preview
Cinematic Storytelling
Create connected shots with deliberate pacing, camera movement, atmosphere, and narrative progression.
Media preview
Product Advertising
Use product references to build reveals, demonstrations, lifestyle scenes, and commercial concepts.
Media preview
Character Performance
Direct expressions, movement, dialogue, voice, and camera behavior across a longer performance.
Media preview
Fashion & Visual Concepts
Develop styled sequences with coordinated subjects, environments, movement, lighting, and camera direction.
Media preview
Creative Previsualization
Test scene blocking, camera ideas, motion, visual language, and narrative structure before production.
Media preview
Branded Social Video
Build complete short-form concepts with a hook, action, product or story beat, and a clear ending.
How to Use Seedance 2.5 on DoMax AI
01
Choose Seedance 2.5
Open the Seedance 2.5 generator and select the creation mode supported by the tool for your intended workflow.
02
Add your prompt and references
Describe what should happen in the scene. Upload supported image, video, or audio references when they communicate the intended subject, motion, visual direction, or sound more clearly than text alone.
03
Configure the available settings
Choose from the duration, aspect ratio, quality, audio, or other controls currently available in the DoMax AI integration. The generator interface should always be treated as the source of truth for available options.
04
Generate and review
Check subject consistency, motion, camera behavior, sound, pacing, and whether the result follows the requested sequence.
05
Refine the result
Adjust the prompt or references one variable at a time. If a strong result only needs a local change or continuation, use an editing or extension workflow when that capability is available instead of rebuilding the entire concept.
Use the available refinement controls
Tips for Better Results
01
Start with a clear visual event rather than a collection of style keywords. Define who or what is in the scene, what changes over time, and how the camera observes that action.
02
For longer videos, think in connected beats. A simple structure such as establishing shot → main action → close-up or transition → final composition is easier to follow than several unrelated events competing for the same timeline.
03
Assign every reference a job. One image might define the exact product, another the environment, a video the camera movement, and an audio file the voice or rhythm. State those relationships clearly in the prompt.
04
When consistency matters, avoid unnecessary changes to clothing, lighting, environment, camera style, and subject identity inside the same generation. Add complexity only when it contributes to the story.
Seedance 2.5 Limitations
Seedance 2.5 provides substantially more control, but AI video generation still involves iteration.
01
Complex physical interactions can remain difficult, especially when several subjects touch, exchange objects, move rapidly, or interact in a crowded scene. Precise anatomy, hands, small typography, exact product geometry, and highly constrained choreography may also require additional generations or stronger reference material.
02
Longer clips create another challenge: every extra action gives the model more information to coordinate. A focused 30-second sequence with a clear progression is generally easier to control than a prompt that attempts to fit many unrelated scenes into the same generation.
03
Reference quality matters as well. Conflicting images, inconsistent characters, incompatible camera examples, or unclear audio instructions can make the target harder to interpret.
Use Seedance 2.5 when its longer storytelling, reference control, native audio, editing, or extension capabilities solve a real creative problem—not only because it has a newer version number.
Seedance 2.5 FAQ
What is Seedance 2.5?
Seedance 2.5 is ByteDance Seed’s latest major Seedance video model as of September 2026. It is an audio-video generation model focused on longer storytelling, multimodal references, precise editing, extension, and more production-oriented creative control.
How long can Seedance 2.5 videos be?
Seedance 2.5 supports video generation up to 30 seconds in a single pass. The model also supports video extension, allowing a successful sequence to continue while carrying forward its subject, environment, visual style, and narrative direction.
Can Seedance 2.5 generate video from images?
Yes. Images can be used as references for subjects, products, characters, environments, composition, style, and other visual details. The prompt can then direct how those elements move and develop over time.
Does Seedance 2.5 support video and audio references?
Yes. Seedance 2.5 supports multimodal reference workflows using images, video, and audio. ByteDance states that a generation can accept up to 30 images, 10 video clips, and 10 audio clips as reference materials.
Can Seedance 2.5 generate audio with video?
Yes. Seedance 2.5 is an audio-video joint generation model, allowing sound and visuals to be generated as part of the same creative process. This can support dialogue, ambience, effects, performance, and other synchronized audiovisual content.
Can Seedance 2.5 edit existing video?
Seedance 2.5 includes more precise video editing capabilities, including time-based targeted changes and advanced workflows such as reference-based editing, camera-perspective adjustments, and green-screen creation. Exact editing options available on DoMax AI should follow the current generator interface.
What is the difference between Seedance 2.5 and Seedance 2.0?
Seedance 2.0 introduced the unified multimodal audio-video workflow. Seedance 2.5 builds on that foundation with up to 30-second single-pass generation, broader reference handling, stronger long-form continuity, and more precise editing and production controls.
Is Seedance 2.5 good for product videos?
Yes. Seedance 2.5 can use product images and other references to guide appearance while the prompt controls movement, camera direction, environment, lighting, and sound. Reference quality and clear instructions are important when exact product details need to remain consistent.
How do I get better results with Seedance 2.5?
Use a clear subject and action, define the visual progression, describe important camera and lighting choices, and give every reference file a specific purpose. For longer clips, use a small number of connected story beats rather than many unrelated actions.
Is Seedance 2.5 always better than faster Seedance models?
No. Seedance 2.5 is most useful when you need longer storytelling, richer references, native audio, editing, extension, or advanced control. Faster or lighter models may be more efficient for rapid drafts, prompt testing, and high-volume experimentation.
Explore More Seedance Models
Seedance 2.5 is the production-focused option in the current Seedance family, but different workflows may benefit from a faster, simpler, or earlier model.
Multimodal creation
Seedance 2.0
A strong general-purpose option for text, image, audio, and video-guided generation.
View modelFaster iteration
Seedance 2.0 Fast
Designed for workflows where rapid testing and creative variation matter.
View modelAudio + video
Seedance 1.5 Pro
Focused on native audiovisual generation, dialogue, performance, and cinematic scenes.
View modelText + image to video
Seedance 1.0
A simpler visual-generation option with stable motion and native multi-shot storytelling.
View modelCreate with Seedance 2.5
Turn your next idea, image, reference, or creative brief into a longer, more controllable AI video with Seedance 2.5.