Recent Advances in Adaptive Sampling and Reconstruction for Monte Carlo Rendering

International audienceMonte Carlo integration is firmly established as the basis for most practical realistic image synthesis algorithms because of its flexibility and generality. But variance, which appears as visually distracting noise in rendered images, is a persistent challenge. Adaptive sampling and reconstruction algorithms reduce variance by controlling the sampling density and aggregating samples in a reconstruction step, possibly over large image regions. In this paper we survey recent advances in this area. We distinguish " a priori " methods that analyze the light transport equations and derive sampling rates and reconstruction filters from this analysis, and " a posteriori " methods that obtain error estimates, usually with statistical techniques, from sets of samples. They typically estimate the errors of several reconstruction filters, and select the best filter locally to minimize error. We discuss advantages and disadvantages of recent state-of-the-art techniques, and provide visual and quantitative comparisons. Some of these techniques are proving useful in real-world applications, and we aim to provide an overview for practitioners and researchers to assess these approaches. In addition, we discuss directions for potential further improvements. Categories and Subject Descriptors (according to ACM CCS): I.3.3 [Computer Graphics]: Picture/Image Generation— Display algorithms Monte Carlo methods are firmly established as the most practical methods for realistic image synthesis by numerically solving the rendering equation. Even simple Monte Carlo rendering algorithms, such as path tracing, come with a number of very desireable properties including unbiasedness, consistency, and applicability to most scene configurations that are relevant in practice. On the other hand, computation times to obtain visually satisfactory results without noticeable noise artifacts are often in the minutes and hours. Therefore , researchers have proposed a wide variety of noise or variance reduction strategies over the years, from different path sampling strategies (importance sampling, bidirectional techniques, Metropolis sampling) to statistical techniques (density estimation, control variates), or signal processing methods (frequency analysis, non-linear filtering), to name the most prominent ones. In this paper, we survey recent advances in adaptive sampling and reconstruction, which have proven very effective and are making Monte Carlo techniques more practical