45 Painting and Visualization

Art in particular painting, has had clear impacts on the style, techniques, and processes of scientific visualization. Artists strive to create visual forms and ideas that are evocative and convey meaning or tell a story. Over time, painters and other artists have developed sophisticated techniques, as well as a finely tuned aesthetic sense, to help accomplish their goals. As visual-ization researchers, we can learn from this body of work to improve our own visual representations. We can study artistic examples to learn what art works and what does not, we can study the visual design process to learn how to design better visualization artifacts, and we can study the pedagogy for training new designers and artists so we can better train visualization experts and better evaluate visualizations. The synergy between art and scientific visualization, whether manifested in collaborative teams, new painting-inspired visualization techniques, or new visualization methodologies, holds great potential for the advancement of scientific visu-alization and discovery. Scientific visualization applications can be loosely divided into two categories: expository and exploratory. In this chapter, we will focus on exploratory applications. Exploratory applications typically represent complicated scientific data as fully as possible so that a scientific user can interactively explore it. Per the scientific method, a scientist gathers data to test a hypothesis , but the binary answer to that test is usually just a beginning (Fig. 45.1). From the data come ideas for the next hypothesis, insights about the scientific area of study, and predictive models upon which further scientific advances can be made. Exploration of increasingly complicated and interrelated data become a means to that end. One of the most complicated types of data that scientists wish to explore and understand comes in the form of multivalued, multidimen-sional fields. There are a number of visualiza-tion application areas that work with this type of data, including fluid dynamics and medicine. These data are difficult to understand because so many variables, or values, are of interest to the scientists. The challenge comes in understanding the correlations and dependencies between all of the values. For example, 2D fluid-flow simulations produce a 2D vector field that is sometimes time-varying. From this field, additional scalar, vector, and tensor fields are often derived, each relating to the others and providing a different view of the whole. Displaying such multivalued data all together is difficult, even in 2D. It requires showing six to ten different values …