Graphics



In graphics, our interests include parallel image synthesis, efficient scan-conversion schemes, model simplification, terrain visualization, atmospheric effect generation, natural feature synthesis and surface synthesis.

Mesh Simplification and Smoothing
Meshes of triangles are used extensively in computer graphics. Large, dense meshes can be a challenge to handle in certain circumstances, though. We have developed simplification methods that can help in such situations. One contribution of our work in mesh simplification is a novel error metric for shape-preserving mesh simplification, the Quasi-Covariance Error Metric (QCEM). The metric can be used to produce simplified meshes of higher quality than when the popular Quadric Error Metric is used, and it is almost as efficient as the Quadric Error Metric.

Mesh smoothing algorithms are useful if meshes are noisy. We have developed mesh smoothing algorithms that are based on torsion spring formulations: an optimization-based algorithm and a Quasi-Covariance Error Metric-based algorithm. The torsion spring formulation can achieve high quality in mesh smoothing-better than the popular Laplacian smoothing, for example.

Improved Line Clipping
We have investigated alternative line clippers that can perform clipping with improved efficiency. As a result, we were able to develop an algorithm that clips line segments against rectangular clipping regions more efficiently than prior line clipping algorithms. Our algorithm is an improvement to the Cohen-Sutherland line clipping algorithm that avoids a major drawback of the Cohen-Sutherland clipping.

Shadow acceleration in ray tracking using cache mechanisms
We have studied methods for optimizing the shadowing process in ray traced scenes using cache mechanisms. In one part of our work, we developed methods that allow reduction of the computational burden of generating shadowed ray traced renderings. We have also considered management of shadowing process cache using schemes based on Least Recently Used (LRU) and First In, First Out (FIFO) strategies.

Cumulus Cloud Rendering
We are interested in methods for synthesizing images of clouds. Our prior work in this area includes developing new cumulus cloud rendering methods. We have considered using volume rendering techniques, surface-based techniques, and procedural techniques. To generate realistic cumulus cloud images, we designed three techniques (based on Perlin Noise, volume ray casting, and Koch Curve generation). In order to evaluate the new techniques objectively, we also introduced evaluation mechanisms. Since texture and shape are two essential factors in rendering clouds realistically, we used texture measures and shape measures as the basis in the evaluation mechanisms.

Damage Interpolation
We have worked in synthesizing models of damaged objects. An outcome of the work was a method that relied on interpolation of input models.

Animation of organ function over time
We are interested in how animative visualization can help users discover organ function over time (especially cardiac function).

Parallel Rendering
Most of our parallel rendering work is described in the visualization section. We have also investigated parallel surface ray-tracking, however.