Title :
Filter Importance Sampling
Author :
Ernst, Manfred ; Stamminger, Marc ; Greiner, Gunther
Author_Institution :
Lehrstuhl fur Graphische Datenverabeitung, Erlangen-Nurnberg Univ., Erlangen
Abstract :
High quality ray tracing requires pixel sampling and filtering for antialiasing, as well as illumination sampling for complex lighting effects. Both problems are well understood and sophisticated sampling techniques are available for each of them. However, it turns out that weighted pixel sampling spoils the benefits of clever illumination sampling, even with the best filter functions. We propose to circumvent this problem by using pixel sampling and filtering based on importance sampling. Although this technique disallows sharing of samples between neighboring pixels, it results in significant noise reduction and improvement of image quality. The method is easy to implement and well suited for parallel, tile based rendering
Keywords :
antialiasing; filtering theory; image reconstruction; image sampling; importance sampling; ray tracing; realistic images; rendering (computer graphics); antialiasing; illumination sampling; image quality; image reconstruction; importance sampling; noise reduction; parallel-tile based rendering; pixel filtering; pixel sampling; ray tracing; Filtering; Filters; Image quality; Image sampling; Lighting; Monte Carlo methods; Noise reduction; Pixel; Ray tracing; Sampling methods; Filtering; Ray Tracing; Sampling;
Conference_Titel :
Interactive Ray Tracing 2006, IEEE Symposium on
Conference_Location :
Salt Lake City, UT
Print_ISBN :
1-4244-0693-5
DOI :
10.1109/RT.2006.280223