The symposium was organised by the Computer Graphics Laboratory, housed within the Computer Science Department, under the guidance of ETH Professor Markus Gross. Both presentation and participants were an international mixture. Of the 44 contributions only five came from ETH Zurich. Researchers from Europe and the USA from renowned universities, such as Stanford, MIT, UNC Chapel Hill or RWTH Aachen, presented their most recent research results on the point. From a total of 44, the 24 best contributions, ascertained by peer review, were presented and chosen for publication.

Points instead of triangle

The highlights of the symposium were the presentations of the invited speakers. Turner Whitted from Microsoft Research counts as one of the pioneers of point-based computer graphics. Together with Marc Levoy he already introduced the point as a possible graphics element in 1985 in an unpublished article and also laid out an entire rendering architecture. In his symposium lecture, "Are Points the Universal Modeling Primitive?" Whitted talked about the past but also about the future of points in graphics. Microsoft Research is working on a new type of graphic chip architecture, which uses the point instead of the triangle as a basic element.

Leonard McMillan, Professor at the University of North Carolina at Chapel Hill, spoke on the “Architecture Futures for Point-Based Representations“. He delineated a very clear, very rapid rendering architecture for point-based geometry. This architecture was able to compensate for the disadvantage of the point that stems from today's ATI or NVDIA graphic chips, which are only laid out for triangular geometry.

At the beginning of the third day of the symposium Professor Leif Kobbelt from RWTH Aachen addressed the participants with his talk entitled "Point-Sampled Shape Representations“. He explained the mathematical characteristics of point models, such as optimal sampling or estimation error of geometric objects. A birds-of-a-feather discussion forum on Pointshop3D (see box) concluded Kobbelt's presentation.

Point-based computer graphics also attracts the interest of multimedia artists. The picture was created entirely with the ETH-Software "Pointshop3D". large

The picture shows a section from an interactively painted 3D object. Numerous paintbrushes were guided using a haptic instrument; the object – in this case the bee – is rotated or moved with a haptic 3D interactive device. Automatic adjustment of the resolution (resampling) makes it possible to draw very fine details. large

150 years experience with graphics

The second day concluded with a podium event on the subject of "Grand Challenges in Computer Graphics". As Professor Henry Fuchs from the UNC Chapel Hill noted, the five researchers on the podium represented 150 years of experience with graphics, which fitted very neatly into ETH's upcoming Jubilee. For the past eight months Fuchs has spent his sabbatical at in the Computer Graphics Laboratory ETH Zurich. Apart from Fuchs the panel included all invited speakers, as well as Professor Nelson Max from the UC Davis.

Fuchs kicked off the podium discussion with a few open questions in the subject matter of point-based computer graphics. A widespread problem in computer graphics is the storage architecture, which has barely changed since the advent of the computer. A further problem lay in the acquisition of 3D models, that produce millions of points that then have to be processed. In addition a mathematically sound theory was needed for the sampling of point-based geometry and the definition of surface topology.

The podium on point-based graphics (from left): Henry Fuchs, Leif Kobbelt, Nelson Max, Leonard McMillan and Turner Whitted. large

Game stations and 3D cameras

The discussion revolved around the "killer application" for point-based graphics. Apart from game stations, 3D cameras and 3D television also belong to such applications, but they also needed to be accepted by the consumer. Game stations today represented ever more complex scenes and objects. This is precisely where the strengths of the point as a simple basic element could be turned to advantage. New types of 3D cameras are able to absorb both colour and information of the scene's geometry and this is precisely the geometry that is needed to make it possible to represent television pictures in 3D. Moreover, point-based graphics could also be used to produce rapid visualisations of prototypes.

There was wide agreement that it should not be so much a question of finding suitable applications for a problem, but the other way round. Solutions, in which point-based methods would bring an advantage, should be found to existing problems.

The first symposium on point-based graphics succeeded in advancing the scientific dialogue in a new, exciting area of computer science and acted as a catalyst in providing scientists with a publication and communication platform. Following this success the symposium is to be continued in coming years, alternating between Europe and the USA.