Title :
Error analysis and improvement of potential tracing algorithm for global illumination
Author :
Zunce, Wei ; Jizhou, Sun
Author_Institution :
Dept. of Comput. Sci., Tianjin Univ., China
Abstract :
Considering the variance bought up by approximation of radiance during the first pass, this paper presents error analysis and an improvement method for the potential tracing algorithm. With a general form of equations for importance sampling based random walks, variance analysis is given and its relationship to the error of radiance is shown. Aiming at error reduction, a computing method of alternatively solving both adjoint equations is proposed, giving a progressive refinement to both radiance and value of the required flux. Analysis of the method demonstrates a promising convergence rate with iterations.
Keywords :
brightness; convergence of numerical methods; error analysis; image denoising; importance sampling; iterative methods; lighting; ray tracing; rendering (computer graphics); Monte Carlo method; convergence rate; error analysis; error reduction; global illumination; importance sampling; iterations; potential tracing algorithm; progressive refinement; radiance approximation; random walks; variance analysis; Analysis of variance; Approximation algorithms; Convergence; Error analysis; Integral equations; Lighting; Monte Carlo methods; Noise reduction; Sampling methods; Sun;
Conference_Titel :
TENCON '02. Proceedings. 2002 IEEE Region 10 Conference on Computers, Communications, Control and Power Engineering
Print_ISBN :
0-7803-7490-8
DOI :
10.1109/TENCON.2002.1181264