Acting as a task

Give each character something to do to someone, not something to feel.

The task [proven]. For every character in frame, including those who only listen, the PERFORMANCE block (block 9 of the prompt; see Video with Seedance → The prompt grammar) gives:

• what they want right now, what stands in the way, and how they try to get it, written as verbs directed at the other person, with what their eyes are doing built in ("studies his face after each excuse to see whether it worked");

• what the character hides;

• one eye-life line (see Performance → Blinks mark thought);

• one involuntary detail: a breath held, a blink that comes late, a tightened jaw.

For dialogue, read the full exchange before you write anything. The scene gets one direction that everyone shares; each character gets a motive of their own.

Why: asking an actor for a result ("be angrier") gets forced work, and the same holds for a model. A playable verb aimed at a partner gets behaviour. Judith Weston calls the first habit "result direction". Higgsfield's own team builds every performance on what each character wants from the other, and showed the same line acted before and after the change.

How a take's acting is judged (the performance point): see Video with Seedance → Take selection and review. The PERFORMANCE slot in a full prompt: see Video with Seedance → Templates A, B and D.

Sources

• Judith Weston, Directing Actors (Film Courage interview; ScriptReader summary); Michael Caine (article); Higgsfield, "How I Make Realistic AI Short Films Using Seedance 2.5" (video); Tigran Mutafyan (the acting-task method).