Learning to Cluster for Rendering with Many Lights
Journal
ACM Transactions on Graphics
Journal Volume
40
Journal Issue
6
Date Issued
2021
Author(s)
Abstract
We present an unbiased online Monte Carlo method for rendering with many lights. Our method adapts both the hierarchical light clustering and the sampling distribution to our collected samples. Designing such a method requires us to make clustering decisions under noisy observation, and making sure that the sampling distribution adapts to our target. Our method is based on two key ideas: a coarse-to-fine clustering scheme that can find good clustering configurations even with noisy samples, and a discrete stochastic successive approximation method that starts from a prior distribution and provably converges to a target distribution. We compare to other state-of-the-art light sampling methods, and show better results both numerically and visually. © 2021 Copyright held by the owner/author(s).
Subjects
Direct illumination; Many-light rendering; Optimization theory; Ray tracing; Reinforcement learning
Other Subjects
Approximation theory; Computation theory; Monte Carlo methods; Site selection; Stochastic systems; Clustering scheme; Clusterings; Coarse to fine; Direct illumination; Many-light rendering; MonteCarlo methods; Noisy observations; Online monte carlo; Optimization theory; Sampling distribution; Reinforcement learning
Type
journal article