Level-of-Detail Adaptive Subdivision Methods for the Coding of Dynamic Mesh Sequences with V-DMC

The emerging Video-based dynamic mesh coding (V-DMC) standard is designed on the concepts of subdivision surfaces and the lifting scheme. Subdivision surfaces generate different levels-of-detail (LoD) that enable to decode and render a dynamic mesh sequence at different resolutions. The initial specifications of V-DMC only supported the use of the midpoint subdivision method, that is applied to a so-called base mesh, which is a decimated version of the input dynamic mesh sequence frame being processed. Geometry details, capturing the difference between the subdivided mesh and the original mesh are added after the mesh subdivision and are called displacements. While the midpoint subdivision has the benefit of a low complexity, it also generates large displacement data as subdivided vertices lie on the respective planes of the base mesh triangles. In this paper, we present the adaptation of three subdivision methods to V-DMC: Loop, Least Square Subdivision Surfaces (LS3), and Butterfly. We show that combining different subdivision methods at different LoDs results in significant compression performance gains.