Generate a scene of a young Japanese woman singing under weeping cherry blossoms using <image_1> as a character reference. Lip-sync with the input audio. No microphone. Multi-shot.
Prompt Details
An android girl malfunctions and proceeds to destroy an evil research lab one after another. A doctor shouting, "Waaah! Stop it! Please stop it!" but the girl doesn't stop. There is a big explosion, and the lab is destroyed.
How to work from this case
- Start with the original prompt and identify which subject, camera, and mood phrases drive the output.
- When iterating, change one variable first: lighting, motion, or emotion.
- If references are involved, adjust framing and movement separately for more stable generations.
Editorial handling note
Public prompt pages show edited case notes instead of copy-ready raw text. High-risk, explicit, and brand- or likeness-sensitive entries are filtered from the public library.
Related Seedance prompts
Photorealistic blockbuster cinema. 16:9. ARRI anamorphic, heavy 35mm grain, no color grade, only practical lighting. References: Pacific Rim, District 9, Mad Max Fury Road. A battle-damaged mech 40 stories tall stands…
Daytime, a slope lined with cherry blossoms (outdoors). She bursts out of the front door with a piece of toast in her mouth. A parody of the classic anime 'I'm late, I'm late!' trope. The camera captures her profile…
Continue browsing Seedance content
After a single prompt, the next move is usually another example from the library or a guide and glossary entry that add method and terminology.
Prompt Library
Return to the prompt library to keep narrowing
Compare structure, media setup, and control patterns across examples.
Browse
Guides
Read guides for the underlying method
Turn isolated examples into reusable workflow patterns.
Browse
Glossary
Use the glossary to complete the control language
Clarify camera, reference, and consistency language before the next iteration.
Browse