Rychlá implementace procedurálních 3D textur

Procedurally generated noise textures play a key part in the world of computer graphics. However, their computation is very time consuming. Generating noise for rendering of photo-realistic scenes may require large amount of hardware resources. Efficient use of these resources is beneficial in the trend of ever-increasing amount of details in the rendered scenes. In this thesis we develop a series of optimized algorithms for the popular Improved Perlin noise and the new Gabor noise. We will employ SIMD features of the modern generation of general purpose processors to achieve efficient use of hardware resources. These algorithms will be integrated into the MentalRay ray-tracing rendering engine and their effectiveness demonstrated in various scenes. Powered by TCPDF (www.tcpdf.org