The problem of approximating a given input mesh with a less complex but geometrically faithful representation is well-established in computer graphics. Given the visual complexity required to create realistic-looking scenes, simplification efforts can be essential to efficient rendering. Level-of-detail representations figure prominently in real-time applications such as virtual reality, terrain modeling, and scientific visualization, and as a result there is significant demand for effective algorithms for mesh simplification.

Numerous such algorithms have been proposed, and generically they can be said to vary widely in approach, efficiency, quality, and generality. Some techniques offer efficient processing but produce simplified meshes which are visually undesirable. Others create more pleasing approximations but are slow and difficult to implement. Some algorithms go to great lengths to preserve the topology of the input mesh while others alter it arbitrarily. Many methods are restricted to or only perform well on manifold surfaces. In practice, which algorithm is best suited to perform a given simplification depends heavily on the characteristics of the input mesh and the desired attributes of the approximation. In this paper, I give a basic overview of the components of some of the most common mesh simplification algorithms and evaluate their strengths and deficiencies. Readers seeking a more detailed survey of the subject are referred to [Garland 1999a] and those looking for a comprehensive treatment might begin with [Luebke et al. 2002].