[HomePage] [Projects] [Papers] [Students]

Nonphotorealistic Visualization

This project investigates the use of artistic techniques for visualizing large, complex, multidimensional datasets. Many painterly styles correspond closely to perceptual features that are detected by the low-level human visual system. Results from research conducted in our laboratory on the use of perception in visualization offer valuable insights on how to harness, measure, control, and apply painterly techniques to represent a high-density information space. The result is an image that looks like a painting. The painting is not a recreation of a real-world scene, rather, brush stroke properties in the painting are varied to capture changes in values stored in an underlying dataset. We refer to this technique as nonphotorealistic visualization.

This project will study three open questions in the context of nonphotorealistic visualization: (1) How can multiple layers of data be accurately and effectively combined in a single display? (2) How can human perception be harnessed to assist with visualization? and (3) Can we construct presentation methods that encourage viewers to explore and investigate the resulting images?

Four studies will be conducted to address our research goals. First, the painterly styles that are best suited to representing information in a multidimensional dataset will be identified. Second, the fundamental strengths and limitations of these styles will be studied using psychophysical experiments. Third, an effective method for modeling and rendering the individual brush strokes that make up a nonphotorealistic visualization will be constructed. Fourth, techniques that allow animation in time and space through a dataset will be designed and implemented.

A final, important consideration is how well experimental results extend to a real-world visualization environment. Domain experts from oceanography, environmental science, and radiology will be asked to use nonphotorealistic visualization techniques to explore and analyze datasets from their individual areas of research. This will provide a method for validating our findings in a practical setting with real users.


Christopher G. Healey (PI)
James T. Enns (PI, University of British Columiba)

Laura Tateosian (PhD candidate)


Last updated Friday, December 20, 2002