AutoMovie API
    Preparing search index...

    Namespace IAutoMovieMotionHumanPerformanceSoft

    A performance's soft props: each solved once from the displayed poses of the people who hold it or whose bodies it lies on, at its domain's own fixed step, and read by film seconds.

    The pass runs after every action is placed. For each step k of the domain's fixedStepSeconds from film second 0 to the performance's end, it reads the displayed bones at second k dt and builds that step's boundary: the world point of every moving anchor (an actor-bone binding on its person's bone, a node binding on a rigid prop of that subject) and the world segment of every body capsule. A hand's grip on the prop (any grip window or hold on it, as HumanPerformanceRead.grips lists them) binds the particle nearest the hand's grip socket when the grip closes, read from the solve as it stands then, to that socket for the window, and releases it after, so the particle falls with the velocity the hand gave it. Two hands on one particle (a handover's shared window) bind it to the hand that closed last, as a rigid prop rides the last grip. The engine solves the whole span once (simulateSoftBodyWithBoundaries with an observer that keeps every step), replaying from step 0 once per grip that picks its particle from the solve, so W grips cost W + 1 solves.

    A read at a film second interpolates linearly between the two steps around it, so a seek costs no solve and two frame rates read the same state at a shared second. The coupling is one-way: the cloth never moves the body (tension enters the hand as a load through load, read from the compliant solver's multipliers).

    IHandAnchor
    ISolved