Control cage in metres, quads only, with face-varying UVs.
Quads as four vertex indices, counterclockwise seen from outside.
Packed control positions [x, y, z, ...], metres.
Surface coordinate of each face corner, aligned with cage.faces.
Active limit equalities: false for control pins, blend controls and a sample made without fitting.
Each patch's grid, in the network's patch order.
Whether each vertex is a pinned ring vertex, kept at its given position.
Largest distance between a fitted vertex's limit point and its target, metres.
Vertex indices of each named ring in loop order.
Packed pre-inverse generating samples [x, y, z, ...], metres, after girth adjustment and control pins: cell-averaged sweep samples, curve samples and blend controls. Only limitTargetMask rows are hard limit equalities; pinned entries are declared control positions, not geometric curve points, and held blend interiors retain their generating controls.
The tier sampled.
A surface network sampled at one tier: a quad control cage ready for
subdivideAutoMovieSurface, with the surface coordinate of every face corner, the per-patch grids and the named rings.Within a tier the cage's connectivity, UV layout and coordinates depend only on the network's patch and count declarations, never on the shapes of its sweeps and curves, so every input of one network yields the same topology and only positions change. Every tier addresses the same coordinates,
(patch, s, t), so a point bound on one tier is found on another bymapAutoMovieSurfaceTier.