Seedance 2.5 Generates Video and Audio Simultaneously for 30-Second Clips
ByteDance?s Seedance 2.5 generates video and audio together, supports clips up to 30 seconds, and accepts multiple visual and audio references.
ByteDance has released its AI video generation model Seedance 2.5. This version creates video and audio at the same time and can generate clips up to 30 seconds long without separately stitching scenes together. Generated videos can also be extended multiple times.
The input options have expanded as well. Users can provide up to 30 images, 10 videos, and 10 audio files as reference materials to design scenes featuring multiple characters and camera angles. Texture, lighting, and skin rendering have been improved, and results can be edited after generation while maintaining the overall style.
For advertising and short-form video production, the key benefit is reducing the need to create and edit separate clips for each scene. However, maximum lengths and input limits may vary depending on the service or delivery method, so users should check the supported capabilities of the product they are using before applying it to actual projects.
Seedance 2.5 was first made available through China’s Jimeng AI and Doubao Pro, while API access through BytePlus ModelArk is planned for a later date.