Unified Shape and SVBRDF Recovery using Differentiable Monte Carlo Rendering

DiCE: The Infinitely Differentiable Monte-Carlo Estimator

The score function estimator is widely used for estimating gradients of stochastic objectives in Stochastic Computation Graphs (SCG), e.g., in reinforcement learning and meta-learning. While deriving the first order gradient estimators by differentiating a surrogate loss (SL) objective is computationally and conceptually simple, using the same approach for higher order gradients is more challen...

Monte Carlo Methods in Rendering

Mathematical Basics of Monte Carlo Rendering Algorithms

The high computational costs associated with global illumination in the field of computer graphics call for effective ways of handling complicated integrals and sums. This paper offers an overview of Monte Carlo methods, stochastic methods of approximating such constructs. Starting with the necessary mathematical basis for understanding Monte Carlo, the paper continues with a survey of sampling...

Efficient Monte Carlo rendering with realistic lenses

In this paper we present a novel approach to simulate image formation for a wide range of real world lenses in the Monte Carlo ray tracing framework. Our approach sidesteps the overhead of tracing rays through a system of lenses and requires no tabulation. To this end we first improve the precision of polynomial optics to closely match ground-truth ray tracing. Second, we show how the Jacobian ...

Recent Advances in Monte Carlo Offline Rendering

The research problems in Monte Carlo rendering have changed over the years. At first, the question was one of efficiency: how can we render an image quickly? Recent research has focused on rendering more complex effects, such as soft shadows and motion blur. There has also been research into the Monte Carlo technique itself; while it is a wellknown numerical technique, many subtleties have been...