ABSTRACT
The rendering of a detailed and complex model that consists of thousands oftriangles looks quite good when the model is near the eye point. The time it takes to render the large number of triangles is well worth the gain in visual quality. However, when the same model is far from the eye point, the detail provided by thousands of triangles is not that noticeable, because the screen-space coverage of the rendered model might be only a handful of pixels. In this situation, the tradeoff in time versus visual quality is not worth it. If the final rendering covers only a handful of pixels, the number of triangles processed should be proportional. This chapter introduces the concept of geometric level of detail (LOD). The amount of work done by the renderer per model per pixel should be as independent of the number of triangles that make up the model as possible.
