首页> 外文期刊>ACM Transactions on Graphics >Selectively Metropolised Monte Carlo light transport simulation
【24h】

Selectively Metropolised Monte Carlo light transport simulation

机译:选择性大都会蒙特卡洛光传输模拟

获取原文
获取原文并翻译 | 示例

摘要

Light transport is a complex problem with many solutions. Practitioners arenow faced with the difficult task of choosing which rendering algorithmto use for any given scene. Simple Monte Carlo methods, such as pathtracing, work well for the majority of lighting scenarios, but introduceexcessive variance when they encounter transport they cannot sample (suchas caustics). More sophisticated rendering algorithms, such as bidirectionalpath tracing, handle a larger class of light transport robustly, but have a highcomputational overhead that makes them inefficient for scenes that are notdominated by difficult transport. The underlying problem is that renderingalgorithms can only be executed indiscriminately on all transport, eventhough they may only offer improvement for a subset of paths. In this paper,we introduce a new scheme for selectively combining different Monte Carlorendering algorithms. We use a simple transport method (e.g. path tracing)as the base, and treat high variance “fireflies” as seeds for a Markov chainthat locally uses a Metropolised version of a more sophisticated transportmethod for exploration, removing the firefly in an unbiased manner.We use aweighting scheme inspired by multiple importance sampling to partition theintegrand into regions the base method can sample well and those it cannot,and only use Metropolis for the latter. This constrains the Markov chain topaths where it offers improvement, and keeps it away from regions alreadyhandled well by the base estimator. Combined with stratified initialization,short chain lengths and careful allocation of samples, this vastly reducesnon-uniform noise and temporal flickering artifacts normally encounteredwith a global application of Metropolis methods. Through careful designchoices, we ensure our algorithm never performs much worse than the base estimator alone, and usually performs significantly better, thereby reducingthe need to experiment with different algorithms for each scene.
机译:轻型运输是许多解决方案中的复杂问题。从业人员现在面临着为任何给定场景选择使用哪种渲染算法的艰巨任务。简单的蒙特卡洛方法(例如路径跟踪)在大多数照明场景中都可以很好地工作,但是当它们遇到无法采样的传输(例如焦散)时,会引入过多的方差。更复杂的渲染算法(例如双向路径跟踪)可以稳健地处理较大种类的光传输,但是具有较高的计算开销,这使它们对于不受传输困难的场景而言效率低下。潜在的问题是,尽管只能对路径的子集提供改进,但渲染算法只能在所有传输中无差别地执行。在本文中,我们介绍了一种新的方案,用于有选择地组合不同的蒙特卡洛伦德算法。我们使用一种简单的运输方法(例如,路径追踪)作为基础,并将高方差“萤火虫”视为马尔可夫链的种子,该马尔可夫链在本地使用大都会版本的更复杂的运输方法进行勘探,以无偏见的方式移走萤火虫。使用受多重重要性采样启发的加权方案将被积分物划分为多个区域,基本方法可以很好地采样,而不能采样的区域,只能将Metropolis用于后者。这将马尔可夫链约束到可以改进的路径上,并使它远离基本估算器已经很好处理的区域。结合分层初始化,较短的链长和精心分配的样本,这极大地减少了在Metropolis方法的全球应用中通常遇到的非均匀噪声和时间闪烁伪像。通过精心的设计选择,我们确保算法不会比单独的基本估计器性能差很多,并且通常性能会好得多,从而减少了对每个场景使用不同算法进行实验的需要。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号