With the fast growing size and dimensionality of scientific datasets, exploring and rendering data features has become an important topic in visualization. Scientific illustrations have been widely used as visual representations in science and engineering because of their capability to display a large amount of information in a relatively succinct manner. This dissertation investigates new efficient rendering algorithms to improve the visual representation qualities of scientific datasets by integrating the effectiveness of scientific; illustrations with visualization techniques. The main contribution is a volume illustration framework that can visualize volumetric datasets efficiently through conveying object features and simulating multiple illustrative styles. Specifically, a stipple rendering algorithm explores a set of feature enhancements to improve the general understanding of scientific datasets; multiple illustrations styles are achieved through non-periodic 3D pattern and texture generation methods based on Wang Cubes: and an example-based approach creates 3D rendering from 2D illustration examples to simulate professional scientific illustrations. This volume illustration framework can be used to explore features from a dataset interactively and express them efficiently. By taking advantage of geometry-based and hardware-accelerated rendering techniques, important features can be highlighted in an illustrative way at an interactive rendering speed, with a small storage overhead and short preprocessing delay.
Ebert, Purdue University.
Off-Campus Purdue Users:
To access this dissertation, please log in to our