Research

Modeling and Visualization of complex Objects

Many applications in urban and environmental planning, in architecture, landscaping, ecology, arts, film and advertisement need complex objects and scenes for various types of visualizations. We develop efficient algorithms for the creation, level-of-detail modeling and display of scenes with billions of surfaces.

Our work concentrates on botanic objects (trees, landscapes). As early as 1996 I founded the companies Greenworks Organic Software and later Xfrog, Inc. to develop programs for plant modeling. The xfrog modeling software is used widely in film and games and was involved in movie productions such as The time machine and Avatar. Together with the Chinese Academy of Sciences (SIAT Shenzhen) we develop new methods  for scanning tree objects and whole landscapes, thus enabling data-driven models to be shown efficiently in computer graphics scenes. 

Sampling and Rendering

Sampling is one of the basic problems of computer graphics. Good sampling strategies allow producing images with reduced or special kinds of noise. For many years it was not clear what characterizes good sampling sets. We develop various methods to produce such sets and finally found a mathematical description that allows characterizing and tuning such sets optimally. Recent developments are index-based schemes that allow us to produce different kinds of noises and at the same time to generate points extremely fast, which is an important requisite for their practical application in rendering programs.

Non-photorealistic Rendering

These techniques aim at creating abstract visual representations for various applications in simulation, visualization, film and arts. In contrast to the traditional rendering paradigm of creating images that are indistinguishable from photographs, non-photorealistic (abstract) rendering allows to vary geometric details, alter shading styles and modify the geometric representation within a single image. The attention of the viewer can be directed to intended parts, representational styles can help her/him understanding relations between objects, to distinguish between existing and planned objects and much more. Dealing with such rendering techniques allows medical and psychological researchers furthermore to explore and understand the working of our visual system. One project in this research area is our drawing robot e-David

Information Visualization

Information Visualization helps to visually explore complex data and are used in combination with data mining techniques. We work in fields such as software visualization to structurally evaluate very large software systems, in visualization for Biology and Chemistry and text visualization. The latter one might sound strange at the first moment since text itself is a visualization. However, large amounts of text such as given for software systems tend to get invisible since they cannot be comprehended by reading any more. Visualization techniques allow assessing such corpora, the linguistic analysis enables systems to detect important themes and topics. Within the BW-FIT Projektverbund we investigated visualization methods for large high-resolution displays. Within the INCIDE centre we develop new visualisation methods for the life sciences.