AutoMovie API
    Preparing search index...

    Build the function that turns a gated camera subject into the framed subject the camera grammar solves from.

    Called once the performance, couplings and object clips are final, so a follow can track a walking actor, a launched prop or a handoff. The hero's frame moves and every coverage angle read subjects through the same function, so an alternate camera frames the beat's subject exactly as the hero does.

    A group is measured as the union of its resolved members' portable part/proxy boxes, using the displayed world frame of driven nodes and formation footprints sampled from their cues. Height and radius are fixed from the opening union and follow updates the base. Below MIN_MEASURED_SUBJECT_HEIGHT, height takes DEFAULT_SUBJECT_HEIGHT.

    Any other subject resolves to one point, from the live resolver when it answers and the staged placements otherwise, and is boxed through nodeSubjectBox from the geometry its model draws, falling back to its rig span and then the stand-in height. Baked world placements move the full box. Actor translations under a staged facing and live bone positions move their framed base by the same offset. A camera subject holds still.

    type IProps = {
        formationMotions: readonly IAutoMovieFormationMotion[];
        models: readonly IAutoMovieModel[] | undefined;
        motions: Readonly<Record<string, IAutoMovieMotion>>;
        objectMotions: readonly IAutoMovieClip[];
        rotations: ReadonlyMap<string, IAutoMovieQuaternion>;
    }
    Index
    formationMotions: readonly IAutoMovieFormationMotion[]

    Borrowed shot-local formation cues. Empty means formations remain at rest; a cued group's follow position changes while its opening height and radius stay fixed.

    models: readonly IAutoMovieModel[] | undefined

    Borrowed compiled model registry, retained by the returned closure. Undefined or an unresolved model uses the node's rig span, then the stand-in height; measured extents are cached per node.

    motions: Readonly<Record<string, IAutoMovieMotion>>

    Borrowed compiled actor motions keyed by node id. A motion with a staged rotation drives the subject's base before a baked world placement starts; otherwise the subject uses its staged point.

    objectMotions: readonly IAutoMovieClip[]

    Borrowed final object clips in producer order. The framing closure shares one sample of their effective world translation, rotation and scale channels per query instant. The actor motion or staged placement supplies the frame before both required channels start.

    rotations: ReadonlyMap<string, IAutoMovieQuaternion>

    Borrowed map of scene-node world-facing quaternions. A node must have a facing here to follow its actor motion; camera subjects have no staged node facing.