seedance-director

To convert plain-text scene descriptions into structured video prompts, provide your text to generate bilingual JSON outputs featuring detailed camera, action, and character interaction instructions for cinematic video production.

Yes, you can generate bilingual video prompts in English and Chinese. The conversion handles action, general, and dialogue scenes by extracting context to provide tailored instructions for camera movement and character actions.

The best way to create cinematic production prompts for Seedance 2.0 is to input plain-text scene descriptions. The Skill analyzes scene context and extracts relevant elements to build structured JSON prompts detailing camera movements, actions, and atmosphere.

You do not need reference images to generate a video prompt, but providing them alongside your plain-text scene description is supported to help guide the cinematic scene breakdown and prompt generation process.

Scene context analysis for video prompt generation works by evaluating plain-text input to extract relevant elements, categorizing the scene type, and outputting detailed instructions for camera, action, and character interactions.