Fakultät Informatik

GPU-Aware Hybrid Terrain Rendering

 Christian Dick,  Jens Krüger*, Rüdiger Westermann

Computer Graphics and Visualization Group, Technische Universität München, Germany
*Interactive Visualization and Data Analysis Group, DFKI Saarbrücken, Germany



We present a hybrid GPU rendering pipeline for high-resolution textured terrain fields that reduces polygon throughput limitations on the GPU. This pipeline uses rasterization and ray-casting in every frame simultaneously to determine eye ray intersections. It allows shifting workloads flexibly between the triangle setup stage and the parallel execution units, depending on the GPU capacities. We employ a hierarchical tiling of the domain and generate for each tile an error-controlled adaptive tessellation of the terrain. In each rendering frame, for each tile the number of triangles representing this tile as well as the number of pixels covered by the tile are determined, and this information is used to select the fastest rendering method on the executing GPU. Compared to pure rasterization or ray-casting, performance gains of a factor between 2 and 4 are demonstrated for high-resolution fields.


The high-resolution geo data of Vorarlberg was kindly provided by the Landesvermessungsamt Feldkirch, Austria.



Matthias Niessner, our new Professor from Stanford University, offers a number of interesting topics for  master theses.


PhD positions on   Computational Fabrication and 3D Printing and  Photorealistic Rendering for Deep Learning and Online Reconstruction are available at the Computer Graphics & Visualization group.