LoveRain1997/video-to-h3-prompt
Skill Codex
Reverse-engineer a reference video into a complete, ready-to-paste MiniMax H3 prompt (T2VA / I2VA / FL2VA / L2VA / Ref2VA) through dense frame sampling, audio onset/card-point cross-validation, causal event-chain reconstruction, film-type → camera-state-machine triage (continuous take / editorial snap-lock /…
not rated 44
+1
changed 1mo ago A 250 tokens
original MIT