Seedance 2.0 accepts natural-language prompts plus up to nine images, three reference videos totaling 15 seconds, and three audio files totaling 15 seconds, with 12 materials maximum. Users can combine modalities, tag uploads with @ references, choose resolution, duration, aspect ratio, and batch size, then generate multiple versions. It can replicate motion, choreography, camera movement, effects, character appearance, scenes, and sound. Editing tools extend clips, merge scenes, replace characters, alter actions, add elements, or remove content while preserving surrounding material. Built-in audio generation supplies context-aware effects and background music, while uploaded tracks can drive beat-synchronized visuals. Outputs are watermark-free, private, and available through web generation or API access, subject to plan credits and supported settings.