Graphical Visionday, Thursday May 24th 2007



09.10Producing High-Quality Polygonal Meshes
Professor Leif Kobbelt, Computer Graphics and Multimedia, RWTH Aachen
Abstract Today the generation of raw 3D models has become quite easy. Typical sources for geometric data are: 3D scanning, CAD system output, reconstructions from images and video and so on. However, while these models usually have a sufficient quality at the first glance, the removal of inconsistencies and other optimizations are still necessary to make these raw models any useful for downstream applications beyond mere display. Besides this basic mesh repair, one would also like to convert unstructured polygonal models into meshes where individual faces are of high quality in terms of aspect ratio and the degrees of freedom (i.e. vertices) are aligned to major geometric features. These are the global and the local aspects of remeshing techniques respectively. In my talk I will present a number of mesh repair and mesh optimization techniques which are numerically robust and sufficiently efficient to process large dataset of realistic input quality.

SESSION II: Sponsored by The Biophotonics Network

10.20GPU-based Reconstruction of Surfaces from Points
Bjarke Jakobsen, IMM, DTU
Abstract Reconstruction of smooth surfaces from point sets is an important problem in many applications since this sort of data often appear in real-life scenarios. This paper presents method for solving this this problem at interactive rates by means of second order energy minimization obtained by solving for the square Laplacian = 0 on a discrete scalar field using an iterative multigrid approach. Every step in the reconstruction process takes place on the GPU. Consequently, the surface is immediately available for any standard volume visualization method utilizing packed volume textures. This eliminates the need for large amounts of texture transfers or a costly polygonization step.
10.55Advanced Optics for Realistic Rendering
Jeppe Revall Frisvad, IMM, DTU
Abstract Many materials, perhaps most often organic materials, are not opaque. When a translucent or semi-transparent object is present in a scene, we refer to it as a participating medium and render it using ray marching with a single scattering term and an approximate diffusion term, a fully blown Monte Carlo ray tracing, or a Bidirectional Surface Scattering Reflectance Distribution Function (a BSSRDF, sometimes called a subsurface scattering computation). No matter how we choose to render it, if we want a realistic appearance, it is crucial that the correct optical properties are available for the material. In this presentation, I will discuss the optical properties of materials. It turns out that these optical properties can be computed from the particle composition of the material. This gives us great advantages. For example, I will show how we can create a model which, given the fat, protein, and vitamin contents of some milk, can predict the appearance of the milk. Or given the mineral and algal contents of water, we can construct a model predicting its appearance. In other words, I will show how advanced optics makes it possible to construct a highly variable appearance model which computes the optical properties that are crucial in realistic image synthesis.


13.00Art Direction in Animation through Pose Space Deformation
Rasmus Tamstorf, Walt Disney Animation Studios
AbstractShapes are one of the key elements in graphic design. A recurring challenge in high-end character animation is therefore the desire to art direct the apparent shape of characters at all times. Pose Space based Deformation (PSD) provides a method to achieve this through high-dimensional interpolation techniques. Unfortunately, interpolation (especially in high dimensions) is fraught with many problems, so this presentation takes a look at a number of these problems and how they have been addressed in order to produce a highly successful production tool.


14.00Advanced D3D10 rendering
Emil Persson, AMD
Abstract With the introduction of Direct3D 10 a new world of opportunities opened up as well as a new set of caveats for developers to take into account. This speech will introduce you to the new features and concepts of the D3D10 API and explore some new and exciting rendering techniques enabled by this API. A couple of new D3D10 samples from AMD will be demonstrated and the ideas behind them explained in detail. Optimizations and pitfalls will also be discussed.


15.00Irradiance Volumes for Real-Time Dynamic Scenes
Rune Vendler, VGC
Abstract Irradiance volumes, in one form or another, have proven a useful tool in the real-time graphics toolbox. While often precomputed and used to improve the quality of lighting on dynamic objects in static environments, other configurations with different tradeoffs are possible. This talk will discuss an irradiance volume representation and processing algorithm built to support dynamic environments and lighting conditions, trading accuracy for performance, but retaining the ability to express global effects.


16.00Lessons Learned Developing Chili con Carnage
Kim Steen Riber, Deadline Games
Abstract Developing AAA games sets a high standard for the technology driving the game. When developing for small platforms like PSP and PS2 there is a large task in getting your code fast enough to hit the target framerate. This talk will describe some of the areas where we focused our efforts to achieve this goal. You will be surprised how much could be gained by tossing advanced log n algorithms in favor of very simple linear algorithms, simply because the constant factor could be reduced significantly by using the cache in a more efficient way. The presentation will also describe some of the tools used to identify performance bottlenecks, both in cpu, gpu and memory, and what was done to eliminate these performance issues.