David E. Breen (California Institute of Technology, Center for Advanced Computing Research, USA): Level Set Models for Computer Graphics

A level set model is a deformable implicit surface that has a volumetric representation. It is defined as an iso-surface, i.e. a level set, of some implicit function \Phi. The surface is deformed by solving a partial differential equation on a regular sampling of \Phi, a volume dataset. Level set methods provide the techniques needed to change the voxel values of the volume in a way that evolves the embedded iso-surface to meet a user-defined goal. Deforming models within a level set framework provides several advantages. By construction, self-intersection cannot occur, which guarantees the generation of physically realizable, simple, closed surfaces. Additionally, level set models easily change topological genus, and are free of the edge connectivity and triangle quality problems associated with mesh models.

In this talk I will describe three applications of level set models for computer graphics. They include 3D metamorphosis, volume segmentation, and surface editing. Level set morphing provides a powerful and flexible approach to making one 3D model change into another. Level set models for segmentation easily fit to complex 3D structures of arbitrary genus. Level set surface editing allows a user to cut, paste and blend different types of models. Level set models are controlled via a speed function F (x, ...), which is the speed of the level set at point x in the direction of the normal to the surface at x. Therefore all the information needed to deform a level set model may be encapsulated in F, providing a straightforward, unified computational framework for creating computer graphics models.

Marc Alexa (Darmstadt University of Technology, Discrete Geometric Modeling Group, Germany): Representing and Rendering Surfaces from Point Samples

In this talk, I will motivate the use of point sets to represent shapes. I discuss two definitions of a smooth manifold surface from a set of points close to the original surface. These surface definitions are based on local maps from differential geometry, which are approximated using locally weighted least squares. The first approach is based on a projection operation, the second approach on a ray-intersection algorithm -- both approaches define the surface as the stationary points of the operations. These representations also immediately generalize to multi-resolution hierarchies. For rendering, I will concentrate on two techniques: Forward projecting the local maps and ray tracing.

Daniel Cohen-Or (School of Computer Science, Tel Aviv University, Israel): From-region Visibility and Ray Space Factorization

Rendering large scenes in real time remains a challenge as the complexity of models keeps growing. Visibility techniques such as occlusion culling can effectively reduce the rendering depth complexity. Methods that compute the visibility from a point are necessarily applied in each frame during rendering. Recently, more attention is devoted to {\em from-region} methods where the computed visibility is valid for a region rather than a single point. These methods take advantage of time and spatial coherence, and the computational cost of the visibility calculations is amortized over consecutive frames. Still, it is desirable to be able to compute from-region visibility on-the-fly, not having to resort to off-line methods that require excessive storage space.

The from-region visibility problem is considered significantly harder than the from-point visibility problem, as it is inherently 4-dimensional. Our occlusion culling method is based on factorizing the 4D visibility problem into horizontal and vertical components. The visibility of the horizontal component is based on a parameterization space, and the visibility of the vertical component is solved by incrementally merging umbrae. The technique is designed so that the horizontal and vertical operations can be efficiently realized
together by common graphics hardware.

Werner Purgathofer and Alexander Wilkie (CG Group, Thechnical University Wien, Austria): Photo-realistic Rendering - where are we?

After about two decades of computer graphics researchers trying to achieve photo-realism in their images as reliably as possible, the recent years have brought a change of efforts towards real-time methods, easy-to-use systems, integration with vision, modeling tools and the like. The quality of images is mostly accepted as sufficient for real world applications, but is that true? There are still numerous problems to be solved, and there is much progress in these areas. Unfortunately, many of these developments happen in isolated systems for the pure purpose of publication, and never make it into commercial software. This talk wants to motivate people to open their eyes to see such activities, and to be aware of the distance we still have to towards photo-realism.

Tosiyasu L. Kunii (IT Institute, Kanazawa Institute of Technology, Japan): Conceptual Visualization and Graphics

Visualization and graphics have been centered around display of objects and phenomena. This talk explores the meaning, potential and future of conceptual visualization and graphics. Its social impacts are also elaborated through case studies.

Hans-Peter Seidel (MPI Informatik, Saarbruecken, Germany): A Framework for the Acquisition, Processing, and Interactive Display of High Quality 3D Models

High quality 3D models are quickly emerging as a new multimedia data type with applications in such diverse areas as e-commerce, online encyclopaedias, or virtual museums, to name just a few. This talk presents new algorithms and techniques, developed by the speaker and others, for the acquisition and real-time interaction with complex textured 3D objects and shows how these results can be seamlessly integrated with previous work into a single framework for the acquisition, processing, and interactive display of high quality 3D models. In addition to pure geometry, such algorithms also have to take into account the texture of an object (which is crucial for a realistic appearance) and its reflectance behavior. The measurement of accurate material properties is an important step towards photorealistic rendering, where both the general surface properties as well as the spatially varying effects of the object are needed. Recent work on the image-based reconstruction of spatially varying BRDFs enables the generation of high quality models of real objects from a sparse set of input data. Efficient use of the capabilities of advanced PC graphics hardware allows for interactive rendering under arbitrary viewing and lighting conditions and realistically reproduces the appearance of the original object.

H. Suzuki, Y. Takarada, S. Takeuchi, I. Kawano, J. Hotta (The University of Tokyo and Elysium Co. LTD.): A Coding System for Viewing CAD Data based on Mesh Subdivision

In Digital Mockup (DMU) systems or collaborative engineering environment, CAD models usually must be polygonized to a large scale polygon and transmitted over the network. However such large polygons are not suitable for distributing the CAD models on the network and for interactive rendering. We introduce a new CAD data coding system based on the mesh subdivision including subdivision surface approximation to CAD surfaces and a method to subdivide analytic surfaces such as cylinders and cones. In this system, it is possible to compress the models in high rate, and to view them efficiently using a LOD technique. We implemented a system and applied it to CAD models.

Myung-Soo Kim, Wenping Wang and Bert Juettler (Seoul National University, Seoul, Korea): Applications of Ellipsoids in Geometric Modeling and Processing