A Machine Learning Approach for Filtering Monte Carlo Noise


Proceedings of SIGGRAPH 2015

ACM Transactions on Graphics (TOG)
Vol. 34, No. 4, August 2015

     Nima Khademi Kalantari          Steve Bako          Pradeep Sen     

University of California, Santa Barbara

Abstract

The most successful approaches for filtering Monte Carlo noise use feature-based filters (e.g., cross-bilateral and cross non-local means filters) that exploit additional scene features such as world positions and shading normals. However, their main challenge is finding the optimal weights for each feature in the filter to reduce noise but preserve scene detail. In this paper, we observe there is a complex relationship between the noisy scene data and the ideal filter parameters, and propose to learn this relationship using a nonlinear regression model. To do this, we use a multilayer perceptron neural network and combine it with a matching filter during both training and testing. To use our framework, we first train it in an offline process on a set of noisy images of scenes with a variety of distributed effects. Then at run-time, the trained network can be used to drive the filter parameters for new scenes to produce filtered images that approximate the ground truth. We demonstrate that our trained network can generate filtered images in only a few seconds that are superior to previous approaches on a wide range of distributed effects such as depth of field, motion blur, area lighting, glossy reflections, and global illumination.

The Video

A Machine Learning Approach for Filtering Monte Carlo Noise

Paper and Additional Materials

Bibtex

@article{LBF,

author = {Nima Khademi Kalantari and Steve Bako and Pradeep Sen},

title = {{A Machine Learning Approach for Filtering Monte Carlo Noise}},

journal = {ACM Transactions on Graphics (TOG) (Proceedings of SIGGRAPH 2015)},

volume = {34},

number = {4},

year = {2015},

}





The information (source code, data sets, etc.) provided on this website is for non-commercial, research purposes only. For commercial inquiries regarding the learning based filtering algorithm presented in this paper, please contact Mr. Christopher Del Vecchio at the UCSB technology and alliances office. His telephone number is (805) 893-5209 and email address is delvecchio@tia.ucsb.edu.

This page contains documents, source code, videos and other files that could be protected by copyright. They are provided here for reasonable academic fair use. Interested parties are referred to the official published version of the documents which are available from the copyright holder through the external link.

This material is based upon work supported by the National Science Foundation under Grants No. IIS-1342931 and IIS-1321168.

Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.