Nvidia Corporation

Disney Enterprises, Inc.

Eidgenossische Technische Hochschule Zurich

Thomas Müller

Fabrice Pierre Armand Rousselle

Alexander Georg Keller

Jan Novák

Jan Novak

Nikolaus Binder

Morgan McGuire

Peter Schuyler Shirley

Brian McWilliams

Ward Lopes

Matthew Milton Pharr

Marios Papas

Wojciech Krzysztof Jarosz

David Augustus Hart

Monte Carlo and quasi-Monte Carlo integration are simple numerical recipes for solving complicated integration problems, such as valuating financial derivatives or synthesizing photorealistic images by light transport simulation. A drawback of a straightforward application of (quasi-)Monte Carlo integration is the relatively slow convergence rate that manifests as high error of Monte Carlo estimators. Neural control variates may be used to reduce error in parametric (quasi-)Monte Carlo integration—providing more accurate solutions in less time. A neural network system has sufficient approximation power for estimating integrals and is efficient to evaluate. The efficiency results from the use of a first neural network that infers the integral of the control variate and using normalizing flows to model a shape of the control variate.

Neural network control variates

A fully-connected neural network may be configured for execution by a processor as a fully-fused neural network by limiting slow global memory accesses to reading and writing inputs to and outputs from the fully-connected neural network. The computational cost of fully-connected neural networks scale quadratically with its width, whereas its memory traffic scales linearly. Modern graphics processing units typically have much greater computational throughput compared with memory bandwidth, so that for narrow, fully-connected neural networks, the linear memory traffic is the bottleneck. The key to improving performance of the fully-connected neural network is to minimize traffic to slow 'global' memory (off-chip memory and high-level caches) and to fully utilize fast on-chip memory (low-level caches, “shared” memory, and registers), which is achieved by the fully-fused approach. A real-time neural radiance caching technique for path-traced global illumination is implemented using the fully-fused neural network for caching scattered radiance components of global illumination.

Fully-fused neural network execution

Monte Carlo and quasi-Monte Carlo integration are simple numerical recipes for solving complicated integration problems, such as valuating financial derivatives or synthesizing photorealistic images by light transport simulation. A drawback of a straightforward application of (quasi-)Monte Carlo integration is the relatively slow convergence rate that manifests as high error of Monte Carlo estimators. Neural control variates may be used to reduce error in parametric (quasi-)Monte Carlo integration&#x2014;providing more accurate solutions in less time. A neural network system has sufficient approximation power for estimating integrals and is efficient to evaluate. The efficiency results from the use of a first neural network that infers the integral of the control variate and using normalizing flows to model a shape of the control variate.

A fully-connected neural network may be configured for execution by a processor as a fully-fused neural network by limiting slow global memory accesses to reading and writing inputs to and outputs from the fully-connected neural network. The computational cost of fully-connected neural networks scale quadratically with its width, whereas its memory traffic scales linearly. Modern graphics processing units typically have much greater computational throughput compared with memory bandwidth, so that for narrow, fully-connected neural networks, the linear memory traffic is the bottleneck. The key to improving performance of the fully-connected neural network is to minimize traffic to slow 'global' memory (off-chip memory and high-level caches) and to fully utilize fast on-chip memory (low-level caches, 'shared' memory, and registers), which is achieved by the fully-fused approach. A real-time neural radiance caching technique for path-traced global illumination is implemented using the fully-fused neural network for caching scattered radiance components of global illumination.

A real-time neural radiance caching technique for path-traced global illumination is implemented using a neural network for caching scattered radiance components of global illumination. The neural (network) radiance cache handles fully dynamic scenes, and makes no assumptions about the camera, lighting, geometry, and materials. In contrast with conventional caching, the data-driven approach sidesteps many difficulties of caching algorithms, such as locating, interpolating, and updating cache points. The neural radiance cache is trained via online learning during rendering. Advantages of the neural radiance cache are noise reduction and real-time performance. Importantly, the runtime overhead and memory footprint of the neural radiance cache are stable and independent of scene complexity.

Real-time neural network radiance caching for path tracing

Sampling a function is used for many applications, such as rendering images. The challenge is how to select the best samples to minimize computations and produce accurate results. An alternative is to use a larger number of samples that may not be carefully selected in an attempt to increase accuracy. For a function that is an integral, such as functions used to render images, a sample distribution may be computed by inverting the integral. Unfortunately, for many integrals, it is neither easy nor practical to compute the inverted integral. Instead, warp functions may be combined to provide a sample distribution that accurately approximates the factors of the product being integrated. Each warp function approximates an inverted term of the product while accounting for the effects of warp functions approximating other factors in the product. The selected warp functions are customized or 'fitted' to implement importance sampling for the approximated product.

Estimating product integrals using a composition of warps

According to one implementation, a system includes a computing platform having a hardware processor and a system memory storing a software code including multiple artificial neural networks (ANNs). The hardware processor executes the software code to partition a multi-dimensional input vector into a first vector data and a second vector data, and to transform the second vector data using a first piecewise-polynomial transformation parameterized by one of the ANNs, based on the first vector data, to produce a transformed second vector data. The hardware processor further executes the software code to transform the first vector data using a second piecewise-polynomial transformation parameterized by another of the ANNs, based on the transformed second vector data, to produce a transformed first vector data, and to determine a multi-dimensional output vector based on an output from the plurality of ANNs.

Piecewise-polynomial coupling layers for warp-predicting neural networks

The disclosure provides an approach for rendering heterogeneous polydisperse granular media. In one aspect, a rendering application renders such granular media using a combination of explicit path tracing and accelerated path construction using proxy path tracing, shell tracing, and volumetric path tracing. In proxy path tracing in particular, the rendering application instantiates proxy geometry in the form of a bounding sphere and determines internal scattering in the grain using a precomputed grain scattering distribution function that relates incident and outgoing radiance functions on the bounding sphere. In shell tracing, the rendering application uses shells to aggregate many grain interactions into a single step. The rendering application derives a continuous liquid/volume equivalent to the granular material based on the material's optical properties and selects a precomputed shell transport function (STF) to use from a database by interpolating nearby database entries using radiative transfer equation (RTE) parameters at the shell's center.

Growing community of inventors

Rheinfelden, Germany

Thomas Müller