NARU API
    Preparing search index...

    Interface BatchResidencyShape

    Batch payload sizes, which accessor counts already determine before a decode.

    interface BatchResidencyShape {
        edgeVertexBytes: number;
        instanceCount: number;
        sharesSurfaceVertices?: boolean;
        surfaceIndexBytes: number;
        surfaceVertexBytes: number;
    }
    Index
    edgeVertexBytes: number
    instanceCount: number
    sharesSurfaceVertices?: boolean

    True when a batch decoded earlier already charges these vertices. Material groups of one prototype share its vertex pool, so only the batch that owns the pool pays for it -- and the renderer allocates no second buffer, so a sharing batch adds nothing, not even an empty buffer's four bytes.

    surfaceIndexBytes: number
    surfaceVertexBytes: number