Introduction

Polygon meshes are currently the most used model representations for visualization of three-dimensional computer graphics. These meshes are mathematically simple to manage and polygon mesh rendering hardware is widely available. Most continous model representations (splines, bezier curves, volume data) can be approximated with arbitrary accuracy.

  

Figure 1: Model of the Cassini space probe (415.000 polygons)

The complexity of models for many applications (CAD, medical or scientific visualizations) is too high to render it with available graphic hardware interactively (appropriate framerate). Figure 1 with 415.000 polygons and figure 2 with 500.000 polygons are examples of such very complex 3D scenes. The scenes of the Michelangelo project contain up to 500 millions triangles and current hardware can not display them properly [4]. Generally, three different approaches for simplifying polygon meshes can be distinguished:

Augmenting:

Additional details are added to the raw polygon data to achieve a better visual quality. Some examples are Gouraud-shading and texture-mapping.

Culling:

Using additional informations about the model, parts of the model are not displayed. These parts can be objects out of view or behind the spectator.

Polygon simplification:

Polygons of small, not important or far objects are simplified or temporally erased without a significant loss in visual quality.

  

Figure 2: Submarine auxiliary room (500.000 polygons)