Models · 11 June 2026 · 1 min read
Seedance 2.0: the model that takes pictures, video and sound
Launched on 12 February 2026: a model that takes up to 15 files of reference and returns 15-second multi-shot video with stereo sound.
By The Crewnema team
ByteDance launched Seedance 2.0 on 12 February 2026. The idea was that one request could combine text, pictures, video and sound, and the model would follow all of them. This is what ByteDance said at launch.
What ByteDance announced
- Four kinds of input: text, images, audio and video, with up to 9 images, 3 video clips and 3 audio clips in one request.
- Output of up to 15 seconds, with several shots in one clip.
- Dual-channel stereo sound with several tracks at once: background music, ambient effects and voiceover.
- Reference generation: composition, motion, camera movement, visual effects and audio can be taken from the files you supply.
What it means for an advert
The shift is from describing what you want to showing it. A photo of the real product, a clip with the camera move you like and a voice you want can all go in together. For a business, the most useful of those is the first: a real photo as a reference is the best way to keep a product looking like itself.
In Crewnema
Seedance 2.0 is the video model at the High quality level, which is the default for a new video. Your product photo and the photos of your people become the references. If a clip needs more detail or a longer take, Seedance 2.5 sits at Max. The wider guide is how to get high-quality AI video.
Sources
Make your first video
Crewnema is in private beta and lets businesses in a few at a time. Request access and we will write when yours is open.