AutoMovie API
    Preparing search index...

    Namespace HumanLimbFitFrame

    The frame a limb fit is solved in (solveHumanLimbFit): the root's world position and heading, so two people of one body doing the same thing in different places, or one body facing another way, solve the same numbers.

    A pose and its goal are moved into the frame by taking away the root's world position on all three axes and turning by minus its heading about +Y; the limb solver reads only the geometry relative to the root, so a fit solved in the frame is the fit solved in the world, moved. The fitted pose is handed back with the caller's own root, since a limb fit changes the limb's joints and not the root; its seed (the swivel and branch of fitSeededChainToTarget) lives in the chain root's bone-local frame, which a turn about +Y and a world translation leave as they are, so it is handed back unchanged. A goal attached to a bone (on), or kept clear of or resting on other bodies' colliders (clear, rests), names world things the frame would have to carry, and is solved in the world; so is a pose without a root.

    The limb solver has no absolute floor-height term: free-leg geometry, own-body arm clearance and contact re-aim depend on relative positions, local contact offsets and rotations. Every root, target, bend and rival point moves together. A world-fixed ground goal remains different when only the root moves vertically, because its relative target Y changes.

    goal
    of
    point
    pose