seedance-prompt-en

Seedance prompts for multimodal video generation are written using a structured reference system that assigns assets like text, images, videos, and audio to specific roles within your prompt. This ensures controlled delivery and high-quality output.

Time-segment prompting structures your video generation instructions into specific intervals like 0–3s and 3–6s. This controls the visual rhythm and delivery, ensuring precise scene transitions and timing in the final output.

Reference images and audio are supported in Seedance 2.0 prompts through the @ reference system. This system correctly assigns multimodal assets to designated roles, specifying their exact purpose within the generated video.

Limitations for generating short drama and ad spot videos include enforced non-face content rules and specific asset input limits. These constraints ensure safety while guiding creative expression within your target duration.

The best way to structure a product ad video prompt is using template-driven workflows provided by the Skill. These templates integrate time-segment prompting, camera descriptors, and the @ reference system for consistent ad spots.

Specifying camera and effect descriptors in video generation prompts is supported to achieve precise visual control. Combining these descriptors with time-segments and constraints ensures your multimodal video matches your creative blueprint.