Matthias Holländer - GPU-Guy : Geometry Processing/Realtime Rendering

Level-of-Detail structures are a key component for scalable rendering. Built from raw 3D data, these structures are often defined as Bounding Volume Hierarchies, providing coarse-to-fine adaptive approximations that are well-adapted for many-view rasterization. Here, the total number of pixels in each view is usually low, while the cost of choosing the appropriate LoDs for each view is high. This task represents a challenge for existing GPU algorithms. We propose ManyLoDs, a new GPU algorithm to efficiently compute many LoDs from a Bounding Volume Hierarchy in parallel by balancing the workload within and among LoDs. Our approach is not specific to a particular rendering technique, can be used on lazy representations such as polygon soups, and can handle dynamic scenes. We apply our method to various many-view rasterization applications, including Instant Radiosity, Point-Based Global Illumination, and reflection/refraction mapping. For each of these, we achieve real-time performance in complex scenes at high resolutions.

Ambient Occlusion (AO) provides an effective approximation to global illumination that enjoys widespread use amongst practitioners. In this paper, we present a fast easy-to-implement separable approximation to screen space ambient occlusion. Computing occlusion first along a single direction and then transporting this occlusion into a second pass that is stochastically evaluating the final shading based on the AO estimated proves extremely effecient. Combined with interleaved sampling and geometry-aware blur, visually convincing results close to a non-separable occlusion can be obtained at much higher performance.

We propose a new GPU method for synthesizing subdivision meshes with exact adaptive geometry in real time. Our GPU kernel builds upon precomputed tables of basis functions for subdivision surfaces and is therefore supporting all subdivision schemes, either interpolating or approximating, for triangle or quad meshes. We designed our kernel so that it can be integrated seamlessly within a standard tessellation pipeline, exploiting software or hardware (adaptive) tessellation methods. We make use of the tessellator unit as an adaptive mesher for maximum subdivision level, exploiting the linear nature of subdivision surfaces to enable arbitrary level of detail adaptivity and control the visual smoothness using Subdivision Shading by applying the same tables as for geometry. We evaluate our kernel on a variety of dynamic meshes and compare it to subdivision substitutes.

In this paper we present the progress of collaborative ongoing work into the realistic 3D animation of a tennis player from the 3DLife project. The main focus of this work revolves around producing a realistic 3D virtual avatar of an athlete, capturing realistic motion data of the athlete via a pre-recorded motion database, streaming the motion over a network and displaying the virtual human model of the athlete in a virtual and fully interactive environment. While we have focused on tennis in this paper we believe our methods can be generalised to a wide range of other activities both in sports and otherwise.

I wrote an 'off' loader for java script. The off file must be wrapped into an xml tag.

This is a tiny Maxscript for exporting an animation to an obj-sequence.
This was used for example in our ManyLoD paper for the character animation.

Recently I experimented a little with C# and converted my C++ Halfedge Data structure to C#. This is not yet the finished version.

Latex file to merge several pdfs. For images you can use Irfan View to convert your images to pdf and then use this file.