| The purpose of scientific computation is insight. Visualization is the most direct way to bring out and deliver insight. The advances of technology have allowed scientists to simulate or observe our physical world at a much higher resolution than was available previously. As a result, large-scale numerical models and high resolution data acquisition devices can produce data in a quantity that traditional visualization methods cannot handle efficiently. The objective of this dissertation is to develop interactive three-dimensional visualization techniques for volume data sets commonly obtained from numerical simulations, such as fluid flow simulations, and data acquisition devices, such as medical imaging devices. Interactivity is being able to change the visualization parameters while watching the changes take effect in real-time on the screen. Interactivity allows features, patterns, and anomalies hidden in the data to be easily captured and thus greatly increases the productivity of scientists. The visualization methods introduced here have been developed by taking advantage of the coherence in the data as well as in the rendering processes, and by taking advantage of parallel processing to achieve highly interactive rendering rates. Furthermore, a physical model has been incorporated into the rendering process to obtain visualization of vector fields that is physically accurate. The highly interactive techniques developed in this research can be integrated into a numerical simulation for monitoring and controlling the modeling process while the simulation is running. |
