Real-Time Volumetric Lighting using SVOs

Detta är en Uppsats för yrkesexamina på avancerad nivå från Lunds universitet/Institutionen för datavetenskap

Sammanfattning: This thesis experiments with the data structure of a sparse voxel octree (SVO)to see if it may improve the performance of empty space ray marching in volumes. While ray marching is a somewhat new technique it is used more often in traversal of volumes. It can be used for realistic volumetric effects in computer games or it can be used in the medical field when examining and visualizing MRI scans. While it has many uses it is however very computationally heavy. The usage in real-time applications is therefore limited as the hardware must be able to maintain enough frames per second to satisfy the standard. Normally one would sample the volume with a fixed sample step in order to extract the information in the volume, even if there is just empty space. The idea of a sparse octree is that it allows the ray to take greater steps past this empty space and thus only sample the actual data. This thesis will explain how to implement and use an octree when rendering smoke in a volume, and showcasing the many challenges that comes with this. The result is compared with a three dimensional texture of the same smoke.

  HÄR KAN DU HÄMTA UPPSATSEN I FULLTEXT. (följ länken till nästa sida)