Kernel-Predicting Convolutional Networks for Denoising Monte Carlo Renderings
We introduce a deep learning approach for denoising Monte Carlo-rendered images that produces high-quality results suitable for production.
July 20, 2017
ACM SIGGRAPH 2017
Steve Bako (University of California Santa Barbara)
Thijs Vogels (Disney Research/ETH Joint M.Sc.)
Brian McWilliams (Disney Research)
Mark Meyer (Pixar Animation Studios)
Jan Novak (Disney Research)
Alex Harvill (Pixar Animation Studios)
Pradeep Sen (University of California, Santa Barbara)
Tony DeRose (Pixar Animation Studios)
Fabrice Rousselle (Disney Research)
Kernel-predicting Convolutional Networks for Denoising Monte Carlo Renderings
We train a convolutional neural network to learn the complex relationship between noisy and reference data across a large set of frames with varying distributed eects from the film Finding Dory (le). The trained network can then be applied to denoise new images from other films with significantly different style and content, such as Cars 3 (right), with production-quality results.
Regression-based algorithms have shown to be good at denoising Monte Carlo (MC) renderings by leveraging its inexpensive by-products (e.g., feature buffers). However, when using higher-order models to handle complex cases, these techniques often overt to noise in the input. For this reason, supervised learning methods have been proposed that train on a large collection of reference examples, but they use explicit filters that limit their denoising ability. To address these problems, we propose a novel, supervised learning approach that allows the filtering kernel to be more complex and general by leveraging a deep convolutional neural network (CNN) architecture. In one embodiment of our framework, the CNN directly predicts the nal denoised pixel value as a highly non-linear combination of the input features. In a second approach, we introduce a novel, kernel-prediction network which uses the CNN to estimate the local weighting kernels used to compute each denoised pixel from its neighbors. We train and evaluate our networks on production data and observe improvements over state-of-the-art MC denoisers, showing that our methods generalize well to a variety of scenes. We conclude by analyzing various components of our architecture and identify areas of further research in deep learning for MC denoising.