Computer graphics algorithms that imitate non-photographic techniques such as painting or pen-and-ink are referred to as non-photorealistic rendering (NPR). An underlying assumption in NPR is that artistic techniques developed by human artists have intrinsic merit based on the evolutionary nature of art. For this reason techniques are usually borrowed directly from artists rather than reinvented from first principles.
The many non-photorealistic techniques used in computer graphics [3,5,6,7,13,14,17,24,28] vary greatly in their level of abstraction. Those that produce a loss of detail, like watercolor or pen-and-ink, produce a high level of abstraction. Several previous papers use a low level of abstraction which preserves precise shape properties and are thus well suited to technical illustration.
Although there is a wealth of computer graphics research dealing with the display of three dimensional images, there has been little exploration into utilizing artistic techniques to aid in the conveyance of shape information in an interactive setting. Kondo et al. proposed a system based on enhancing shape recognition . More recently Mochizuki et al. built a system that aids shape comprehension by rendering edge lines. Markosian et al.  developed algorithms for probabilistically calculating only the silhouettes for polyhedral models in real time. Rossignac et al. used the frame buffer to to render lines that enhance shape recognition . There are also 3D paint programs which allow the user to experiment with non-photorealistic methods [12,25] but these methods restrict interaction and require users trained in traditional drawing and painting techniques. In contrast, our methods incorporate user interaction with automatically generated 3D technical illustrations based on geometric models.