Markov Chain Monte Carlo (MCMC) rendering is extensively studied, yet itremains largely unused in practice.We propose solutions to several practicabilityissues, opening up path space MCMC to become an adaptive samplingframework around established Monte Carlo (MC) techniques. We addressnon-uniform image quality by deriving an analytic target function for imagespacesample stratification. The function is based on a novel connectionbetween variance and path differentials, allowing analytic variance estimatesfor MC samples, with potential uses in other adaptive algorithms outsideMCMC. We simplify these estimates down to simple expressions using onlyquantities known in any MC renderer. We also address the issue that mostexisting MCMC renderers rely on bi-directional path tracing and reciprocaltransport, which can be too costly and/or too complex in practice. Instead,we apply our theoretical framework to optimize an adaptive MCMC algorithmthat only uses forward path construction. Notably, we construct ouralgorithm by adapting (with minimal changes) a full-featured path tracer into a single-path state space Markov Chain, bridging another gap betweenMCMC and existing MC techniques.
展开▼
机译：Markov Chain Monte Carlo（MCMC）渲染广泛研究，但它在很大程度上在实践中仍未使用。我们提出了解决方案的解决方案问题，打开路径空间MCMC成为自适应采样框架围绕已建立的蒙特卡罗（MC）技术。我们地址通过导出图像空间的分析目标函数来实现非均匀图像质量样本分层。该功能基于新颖的连接在方差和路径差异之间，允许分析方差估计对于MC样本，在外面的其他自适应算法中具有潜在用途MCMC。我们将这些估计简化为仅使用简单的表达式任何MC渲染器中已知的量。我们还解决了最多的问题现有的MCMC渲染器依赖于双向路径跟踪和互动运输，在实践中可以太昂贵和/或太复杂。反而，我们应用我们的理论框架来优化自适应MCMC算法这只使用前向路径建设。值得注意的是，我们构建我们的通过调整（最小变化）将全功能路径示踪剂调整为单路径状态空间Markov链，桥接另一个间隙之间的算法MCMC和现有MC技术。
展开▼
