Practical Application of Local Estimations: Singularity Removal in the Solution of the Global Illumination Equation
The work is devoted to the elimination of the singularity in the local-estimation-based solution of the global illumination equation. The presence of the singularity prevents the practical application of this method and the creation of software based on it. A method is proposed to eliminate this problem by averaging the sought value in a certain neighbourhood of the point of interest. The transition to averaging occurs only at small distances between the collision point and the point of interest. In other cases, the calculation is conducted according to the usual local estimation algorithm. The results of the light distribution modelling in the modified Cornell box scene are presented as visualizations for different values of the number of rays, threshold of transition to averaging and the averaging sphere (neighbourhood) radius. The proposed method and the results obtained open the way for the practical program realization of local estimations.
Global illumination equation, Monte-Carlo, local estimations, singularity