Choosing the Level of Geometry Complexity

The level of geometric complexity used to model an object should be appropriate for how visible it will be in the scene and how complex the actual object is.

An appropriate level of geometry complexity must be used for each object or portion of an object. Here are some examples of wasteful usage:
  • using a large number of polygons on an object that will never cover more than a small area of the screen
  • using polygons for detail that will never be seen due to camera angle, or culling
  • using large numbers of primitives for objects that may be drawn with fewer primitives. For example, using hundreds of polygons to render a single quad.

Shader techniques such as bump mapping should be considered to minimise geometry complexity, but still maintain a high level of perceived detail. Other techniques such as Level of Detail should also be used. This is particularly true for things such as reflection passes where higher amounts of geometry may not be visible.