Generating a multi-resolution representation is an important step for rendering software packages that produce fast visualizations of huge tetrahedral meshes. This paper presents “a meshing and rendering framework that simplifies large tetrahedral meshes and thereby builds up a multi-resolution model.” The obtained data structure “can be used at run-time to visualize the mesh up to a user-controllable error.”

An adaptive multi-resolution approach and a recursive point-based rendering algorithm are proposed, and are detailed in the course of a well-structured presentation. The following issues are covered: the state of the art in simplification, multi-resolution representation, and volume visualization; the new multi-resolution representation created by progressive simplification operations; vertex hierarchy management; the view-dependent meshing algorithm; and the new point rendering algorithm. The references are recent and adequate.

The proposed framework was tested on five datasets of different sizes, and results differ from dataset to dataset. The authors observe that the point rendering method is much faster and robust, but “it does not have the visual quality as a direct volume renderer.” Both direct volume and point rendering are supported by the proposed volume rendering environment.

In my opinion, the authors fulfill their objective of interactively rendering moderate (about six million tetrahedra) tetrahedral meshes by considering the best available approaches, and proposing both an adaptive data structure and a new rendering algorithm.