Seedance 2.0 prompt shares facial performance

This is the useful part of AI video: not bigger spectacle, but tighter direction. The prompt reads like an animation brief, which is exactly how these models become production tools.

• –FACS plus valence-arousal gives the generator a structured emotional target instead of vague “expressive” language
• –Fixing identity, hair, proportions, age, and styling tackles the biggest failure mode in character video: drift
• –The 15-second format fits short performance studies where micro-expression and continuity matter more than scene complexity
• –Seedance 2.0’s reference-heavy workflow looks well matched to this kind of prompt engineering
• –Posts like this are how model capabilities get operationalized for creators, not just showcased in marketing demos