Adaptive Netzvereinfachung auf der Basis des Quadrik-Fehlermaßes

In computer graphics, more and more detailed models are being used to increase the display quality. Depending on the application, the desired level of detail is, however, different so that the models should preferably be automatically simplified in order to save processor performance or memory space. This simplification of the triangle mesh under consideration should optically reproduce the original mesh as effectively as possible. An extensive comparison of existing methods led to the choice of the quadric error metric as the metric used in the implementation for the error arising during simplification. This metric, described in detail here, is characterized by its generality, high quality and speed. The simplification is performed by continuous edge contraction that evaluates the geometric error via the quadratic distance of the mesh nodes of the simplified mesh to the planes through the corresponding triangular areas of the original model. The application of the implemented method in virtual reality environments with tactile input and output devices also requires an adaptable level of detail in order to minimize the latency between user action and system reaction to such an extent that the immersive impression of the simulation is retained for all supported sensory organs. To this end, continuous mesh implification is reversibly performed and a progressive mesh representation created thus enabling a rapid automatic adaptation of the level of detail to the boundary conditions of the simulation such as the frame refresh or force update rate