Beta Distribution Shadow Mapping for Light Bleeding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Variance shadow mapping methods suffer from 'light bleeding' artifacts due to overly conservative analytical approximations that fail to accurately depict transitions between full illumination and shadow, leading to unrealistic and distorted illumination in computer-generated images.
Innovation Solution
The method employs a normalized incomplete beta function algorithm, utilizing the mean, variance, minimum, and maximum depth values to compute a more accurate illumination intensity value Q, which eliminates light bleeding by considering the entire distribution of depth values within a filter region.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If percentage-closer filtering is used to reduce aliasing in shadow maps, then shadow accuracy is improved, but computational cost and storage requirements increase exponentially
Solution Approach 1:
The patent changes the mathematical parameters used in shadow map filtering from binary step functions to probability density functions (PDFs) characterized by mean and variance. This allows the system to represent depth distributions more accurately without requiring excessive filter region sizes, thereby reducing computational cost while maintaining shadow accuracy.
Solution Approach 2:
Instead of using large filter regions that examine many pixels (excessive action), the patent achieves accurate shadow mapping by using a smaller filter region size combined with probability-based depth estimation. This partial action approach reduces the number of pixels examined while still producing high-quality anti-aliased shadows.
2Measurement precision
If deep shadow maps store visibility functions with multiple depth values, then shadow accuracy is improved, but memory storage requirements increase
Solution Approach 1:
The patent represents depth distributions using probability density functions parameterized by mean and variance, which are compact representations compared to storing multiple individual depth values. This parameterization reduces memory requirements while maintaining the ability to accurately represent depth distributions for high-quality shadow mapping.
3Speed
If variance shadow mapping uses analytical approximations to compute illumination, then processing speed is improved, but light bleeding artifacts occur
Solution Approach 1:
The patent changes the analytical approximation from using only mean and variance to using mean, variance, minimum, and maximum depth values. This enhanced parameter set allows the probability density function to more accurately model the true depth distribution, eliminating light bleeding artifacts while maintaining fast processing speeds through analytical computation.
Data Source
AI summary
A computer-based system forms a more accurate intensity value for virtual light reflected from computer-generated images. The reflected light intensity from the images is computed pixel by pixel using the mean and variance of the depth values of a filter region associated with the pixel, as well as the maximum and minimum values of the depth values in the filter region. The computations use the normalized, incomplete beta function algorithm.