The objective of polygon reduction algorithm is to take a highly detailed geometry and remove polygons while (hopefully) maintaining its overall form.

Polygon reduction has a couple of important uses, related to improving the performace of a geometry. One is that it can be used to create levels of detail and the other is that it can be used to take large innefficient geometries and (hopefully) reduce them to a manageable size without sacrificing any significant detail. Both are related since the measure of 'significant detail' is often a function of viewing distance. However, in many cases there is a lot of 'detail' in an object that is simply unnecessary and contributes nothing to the actual appearance of the model.

Shading plays a significant purpose in helping us to keep the polygon count low. A rounded object with smooth shading will often look much nicer (and use far fewer polygons) than a similar flat-shaded object that attempts to use a large number of polygons to approximate roundness.

Most polygon reduction algorithms attempt to collapse edges using various criteria to determine the order in which the edges should be collapsed. By repeating the process many times on a model, it can be reduced to any target level. A couple of other techniques will be mentioned in the section on level of detail too.

Algorithms that are commonly used (often in combination) include:

• Remove smallest triangles
• Remove smallest edges
• Merge parallel edges

Good polygon reduction tools/algorithms will take into account colour boundaries and texture maps in addition to attempting to maintain the overall form of the geometry.

Example of a 50% reduction: