AutoMovie API
    Preparing search index...

    Variable buildSpaceObjectConst

    buildSpaceObject: (space: IAutoMovieSpace) => THREE.Group = ...

    Build the structural ground of an IAutoMovieSpace: one Mesh per standable surface, grouped under SPACE_GROUP_NAME.

    This is what closes the gap between the space the feet obey and the world the guide passes draw (#1173). The surfaces were already the engine's semantic ground (heightAt, support contacts, walkability) but nothing ever drew them, so a depth or mask pass of a staged scene showed actors floating in a void. Building them as real meshes is enough: every structural pass collects geometry by traversing the scene and checking isMesh, then replaces its material, so the ground joins depth, mask, normal, and outline with no pass-side special case, unlike a GridHelper, which is a LineSegments and must be hidden before every structural pass.

    A patch with an authored opaque color uses the same mesh in the beauty and guide passes. An uncolored support patch writes neither color nor depth in beauty, leaving any independently built floor to supply its appearance. All patches remain structural guide geometry when a guide replaces the material.

    The engine decomposes each footprint into its actual convex pieces, including notches and holes. Floors and ramps fan those pieces; heightfields clip them to authored lattice cells. Vertex heights use the grounding height query and triangles face +Y. Inside a twisted bilinear cell the drawn planar triangles approximate the height rule; identical vertex heights do not prove identical interpolated interior heights.

    A degenerate footprint (fewer than three non-collinear points, rejected by validateSpace, but a hand-built space may still carry one) encloses no area and contributes no mesh rather than an invalid geometry.

    Type Declaration