Seedance 2.5 Online

Seedance 2.5 generates cinematic video from text prompts and image, video, and audio references. Creators can combine up to 50 references, direct camera movement, edit or extend existing shots with plain-language instructions, and generate continuous takes up to 30 seconds with synced audio. Output options include resolutions up to 1080p and landscape, portrait, or square formats. The product is presented for social creators, filmmakers, marketing and advertising teams, e-commerce brands, educators, and other content creators. Its workflow is to add references, describe the shot, then generate and download the video. New accounts receive free credits; additional use is available through credit packs or a subscription.

Seedance 2.5 is a multimodal video generation model that turns text prompts and image, video, or audio references into cinematic video. It creates continuous shots up to 30 seconds, with camera direction and synced audio generated in the same pass. Users can also describe edits in plain language to restyle a shot, replace an element, or extend a take. It addresses the need for coherent video clips without stitching together multiple short fragments or creating a separate sound pass. Users can start with a text prompt alone or upload images, videos, and audio as references, combining up to 50 files. They describe the subject, action, and camera movement, then generate a clip in the panel. The output can be 480p, 720p, or 1080p, with supported aspect ratios including landscape, portrait, and square formats. Users can download the generated clip. For existing videos, they can upload a clip and describe edits such as restyling, replacing an element, or extending a take. The page identifies social creators, influencers, marketing and advertising teams, filmmakers, studios, YouTubers, e-commerce brands, and educators as users. It highlights creators who need coherent clips, synced audio, or several variations without a separate shoot or sound pass. Filmmakers can use references to keep cast and location consistent while iterating on shots, while brands can create product clips and variants from existing images.