@newtake_kr DIALOGUE MAP:
Inner voice → lips closed
Spoken dialogue → speaker only
Listener → silent facial reaction
Off-screen voice → no visible mouth movement
The camera must also know who owns each line.
Five lines are inner thoughts.
Two lines are spoken aloud.
Only two should move the lips.
AI dialogue becomes believable when voice, mouth and camera are mapped separately—not when every line is thrown into one prompt.
@newtake_kr#newtake#newtakecrew