Seedance 2.5 AI Video Generator
Create coherent videos up to 30 seconds from text, images, video, and audio, with support for up to 50 multimodal references. Seedance 2.5 is now live on Shotyard.
Create Longer, More Controlled Videos
Create Longer Continuous Videos
Generate up to 30 seconds in one pass, giving actions and scene changes more time to develop naturally.
Guide Results with More References
Use up to 50 multimodal assets to communicate subjects, environments, motion, sound, and visual style.
Edit Specific Parts of a Video
Upload the original video in Reference to Video and describe the subject, object, or region to change. Seedance 2.5 regenerates the shot while keeping the surrounding motion, composition, and lighting recognizable.
Seedance 2.5 Key Capabilities
- •Native 30-Second Output: Build a longer continuous sequence without assembling several short generations.
- •Up to 50 References: Use a wider mix of image, video, audio, and visual references in one generation.
- •Targeted Video Editing: Use the original video as a reference, then describe the selected subject or region to change in the prompt.
- •Multimodal Direction: Use different media types to communicate appearance, motion, timing, and atmosphere.
- •Consistent Longer Sequences: Keep subjects, lighting, environments, and movement more consistent across a longer video.
Generate Up to 30 Seconds at Once
A native 30-second generation can show a longer action or sequence in one pass. This reduces the need to join several short clips and gives movements, transitions, and scene changes more time to develop.
Combine Multiple Types of References
Multiple references can guide different parts of one result. This published example assigns the opening frame, camera movement, and surrounding areas to specific files. The prompt and output remain paired exactly as shown in the source demonstration.
| Reference Images | Reference Video | Prompt | Generated Video |
|---|---|---|---|
    | Use @Image 1 as the first frame of the scene. Adopt a first-person perspective and refer to the camera movement effect in @Video 1. The upper scene should be based on @Image 2, the left scene on @Image 3, and the right scene on @Image 4. |
Edit Selected Parts of a Video
To revise a specific area, choose Reference to Video, upload the original clip as a reference, and describe what to add, remove, or replace. Seedance 2.5 generates a revised clip guided by the source, helping preserve the surrounding framing and movement while changing the requested detail.
| Input Video | Regenerated Video |
|---|---|
How to Create with Seedance 2.5
Turn a clear idea and a set of useful references into a longer, more controlled AI video.
Add Your References
Collect images, videos, audio, and other materials that show the subjects, motion, sound, or style you want.
Describe the Full Sequence
Describe how the video begins, what happens next, how the camera moves, and how the sequence should end.
Generate and Refine
Keep subjects, lighting, environments, and movement more consistent across a longer video.
Seedance 2.5 Model Specifications
Create videos up to 30 seconds with up to 50 image, video, and audio references using Seedance 2.5 on Shotyard.
| Developer | ByteDance |
| Native clip length | Up to 30 seconds |
| Reference capacity | Up to 50 multimodal assets |
Last updated: July 28, 2026
Seedance 2.5 Questions and Answers
Create videos up to 30 seconds with up to 50 image, video, and audio references using Seedance 2.5 on Shotyard.
What is Seedance 2.5?
Create videos up to 30 seconds with up to 50 image, video, and audio references using Seedance 2.5 on Shotyard.
What are the main Seedance 2.5 capabilities?
Create coherent videos up to 30 seconds from text, images, video, and audio, with support for up to 50 multimodal references. Seedance 2.5 is now live on Shotyard.
How long can Seedance 2.5 videos be?
Seedance 2.5 can generate a native clip up to 30 seconds long. The longer duration gives actions, camera movement, and scene changes more room to develop within one continuous result.
What does the 50-reference capacity include?
Seedance 2.5 supports up to 50 multimodal reference assets. Images, videos, audio, and other visual references can guide subjects, environments, movement, timing, and atmosphere.
How can I change a specific part of a video?
Choose Reference to Video, upload the original clip as a reference, and describe the subject, object, or region to change. Seedance 2.5 then generates a revised video guided by the source clip.
How does Seedance 2.5 improve consistency?
Generate up to 30 seconds in one pass, giving actions and scene changes more time to develop naturally.
What types of videos can I create with Seedance 2.5?
Generate up to 30 seconds in one pass, giving actions and scene changes more time to develop naturally.
How should I prepare a Seedance 2.5 prompt?
Identify each reference clearly, describe the main action and camera behavior, and state what should remain consistent. For a longer video, add a simple timeline for the beginning, middle, and end.