Abstract
The vast capabilities available in scientific computing today enable scientists to simulate phenomena with ever increasing the size and complexity to the point where visualization capabilities have had a difficult time keeping pace. Many approaches have attempted to address this problem in one way or another. Almost without exception, these approaches begin in the standard toolbox, simply selecting an orthographic and perspective camera model for visualizing the data. It has become increasingly difficult however to comprehensively represent these vast datasets within a single image using one of these conventional camera models because of complex occlusions and large variation in the scale of structures of interest.

In this talk, I will discuss my past and present research in camera model design, a new problem solving paradigm in visualization which advocates for designing camera models that best suit a particular application and optimize dynamically according to the data currently being sampled. While the conventional camera models are rigid and have a limited capability, camera model design is a flexible framework for creating cameras that produce images with multiple viewpoints and variable sampling rates. The images created with camera model design carry many of the benefits of those created using conventional cameras as well. The images generated using camera model design have a single layer, are continuous and non-redundant, and are computed efficiently enabling interactive visualizations.