Seedance 2.5

Seedance 2.5 is a next-generation audio and video generation model designed to produce complete, cinematic clips in a single pass. Create continuous videos up to 30 seconds long—one unbroken take, one prompt—without stitching.

Control the look and storyline with up to 50 multimodal reference assets, including text, images, audio, video, and untextured 3D models. Seedance 2.5 supports targeted, precise edits that refine specific parts of what you generated instead of forcing a full restart.

Use it for second-by-second pacing and camera-language direction, from tight product moments to narrative sequences and film pre-visualization. Choose a generation mode, add only the controls you need, then generate MP4 outputs after setting aspect ratio, resolution, duration, and whether to include audio.

What is Seedance 2.5 used for?

Seedance 2.5 helps you generate consistent, controllable AI videos from a prompt, using optional references and (optionally) audio—ideal for cinematic short sequences, ad-style clips, product demos, and pre-visualization.

How long can videos be generated?

You can generate videos up to 30 seconds in a single pass, designed for one continuous take (no stitching).

What kinds of references can I use?

Media mode supports up to 50 total reference assets, including text, images, videos, audio, and untextured 3D models. References can cover subjects, actions, styles, and sounds.

Does Seedance 2.5 support video editing?

It supports deep, controllable editing—keeping the strongest parts of a generated scene while refining only what needs work (for example, changing clothing, repairing objects, redesigning the environment, or removing elements).

What aspect ratios and resolutions are available in the creator?

You can choose from 1:1, 21:9, 4:3, 3:4, 16:9, and 9:16, with output options for 480p, 720p, or 1080p.

What file format do I get after generation?

The creator provides an MP4 download when generation succeeds.