My research activities are mainly in the field of Human-Computer Interaction (HCI) and there has been an emphasis on Image Synthesis and some related work on Image Analysis. Specific fields include: Image Compression, Interactive Graphics Algorithms, Programming Paradigms for Graphics, and Co-operative Data Visualization. My research methodology is, with some exceptions, experimental: theoretical analysis has an important role but the proof of success derives from experimentation with a prototype system.
Since returning to South Africa I have been spending two months at the Dutch National Centre for Computer Science and Mathematics (CWI, Amsterdam). My position there has been that of research advisor. There are also regular exchanges of students between the institutions.
Work at the CWI on a new architecture for raster graphics has produced concrete results in terms of the first prototype system. My contribution has remained primarily that of developing algorithms that make the new hardware viable. Such algorithms are essential because the hardware is such a radical departure from traditional architectures (for example it has no frame buffer) [13, 8, 12]. The most significant contribution in this respect was the development of an angular interpolation method for real-time Phong shading. Phong shading has been a standard technique in graphics but has eluded effective real-time implementations, our architecture via this new technique does achieve this [2]
After constructing the new architecture I observed that it was essentially a difference engine (shades of Babbage). This in turn suggested to me the possibility of using it for the real-time decompression of wavelet compressed images. The initial suggestion was largely confirmed by the research in image analysis conducted together with my masters student P. Marais [4, 5, 15, 16].
My work as a graphics researcher has also resulted in a number of paper on image synthesis [11, 9]. In these cases the experimental evaluation of the ideas was done by honours or masters students and I believe it reflects well on the quality of our work that these publications were accepted by the premier European (and one of the leading world) conferences in the field. These conferences are highly competitive.
Finally there is the work on programming paradigms for computer graphics. This field results in theoretical advances of computer science that will benefit applied fields such as computer graphics. It has long been a contention of mine that detailed examination of the needs of the applied areas of computer science poses the ``hard'' questions that theory has to address. In this case the existing paradigms of computer programming (namely, imperative programming such as object-oriented programming and declarative programming such as constraint based approaches) are all inadequate in very key respects. This realization grew from my PhD work and goes back to what is probably my most cited paper namely [1]. In 1990 Peter Wisskirchen and I established the Eurographics workshop series on Object-Oriented Graphics. This workshop series has been very successful and recently was renamed the Workshop on the Programming Paradigms in Graphics. I have been chair of several of the workshops, edited a number of books and been invited as a speaker to a number of international events on this topic. The books include [7, 6, 14, 17].
Recently I published a paper on the whole question of appropriate paradigms for graphics [3]. My colleague Prof Goosen was largely responsible for the experimental work while I generally performed the theoretical analysis. Other papers have been more specifically on the topic of the integration of constraint-based and object-oriented paradigms [10, 18].