This workshop aims to demonstrate a new way of manipulating SOPs in TOP space, while preserving optimized performances, by leveraging the GPU through geometry shaders to convert an SOP into, or from, a TOP. The geometry shader will be presented and explained first through a series of simple and practical examples, to illustrate its unique capabilities in comparison to vertex and pixel shaders, and how they may be used to accomplish our goal.
Using this knowledge, we will build a GPU accelerated tox responsible for converting an animated SOP into a series of TOPs, storing vertices attributes and connectivity that vary over time. Then another component will use those very TOPs to reconstruct the geometry by using another geometry shader.
Finally, we will explore how to generate, modify, and combine different geometries in TOP, leveraging the power of pixel shaders to do so. Some tools will be presented that show how to apply this technique to point cloud data or kinect recordings.