Image Rendering Using Ray-tracing on Graphics Card

Abstract

\EnglishAbstract{The focus of this thesis is to optimization of the ray tracing method for parallel execution on graphics hardware. I describe the motivation, the principles and the algorithm of the image rendering based on the ray tracing method. The work presents major disadvantages of the ray tracing method, especially its requirement for the computation power. It describes the evolution of a graphics processing unit (GPU) and its suitability for the implementation of ray tracing algorithm. In this paper I compare the ray tracing method with the standard rasterization method. I also provide description of the OpenCL framework, which is standard for programming parallel applications on heterogeneous platforms. I implement the ray tracer application in the OpenCL. Several performance optimizations were applied to the solution related to the graphics hardware. I carried out the benchmark of the implemented ray tracer on the sample scenes for both platforms, the central processing unit (CPU) and the graphics processing unit (GPU). The ray tracer performance was also analysed via the profiling technique. The main conclusion drawn from this work (analysis) is that the GPU is not a suitable hardware for rendering based on the ray-tracing method, because the benchmark result was insufficient.