Sampling & Filtering

Technical Papers

Sampling & Filtering

Wednesday, 12 August 3:45 PM - 5:35 PM | Los Angeles Convention Center, Room 152 Session Chair: Jaroslav Křivánek , Charles University in Prague


Adaptive Rendering Based on Weighted Local Regression

A novel, weighted local regression based on ab adaptive reconstruction and sampling technique to effectively handle a wide variety of MC rendering effects.

Bochang Moon
Korea Advanced Institute of Science and Technology

Nathan Carr
Adobe Systems Incorporated

Sung-Eui Yoon
Korea Advanced Institute of Science and Technology

Adaptive Rendering With Linear Predictions

This paper proposes a new adaptive rendering algorithm that enhances the performance of Monte Carlo ray tracing by reducing the noise (variance) while preserving a variety of high-frequency edges in rendered images through a novel prediction-based reconstruction.

Bochang Moon
Disney Research Zürich

José A. Iglesias-Guitián
Disney Research Zürich

Sung-Eui Yoon
Korea Advanced Institute of Science and Technology

Kenny Mitchell
Disney Research Zürich

A Machine-Learning Approach for Filtering Monte Carlo Noise

In this machine-learning approach to filter Monte Carlo noise, a neural network is trained in combination with a filter on a set of scenes with a variety of distributed effects and tested on a set of different scenes. The result is improvement over existing approaches.

Nima Khademi Kalantari
University of California, Santa Barbara

Steve Bako
University of California, Santa Barbara

Pradeep Sen
University of California, Santa Barbara

Gradient-Domain Path Tracing

This new path-tracing algorithm, which improves squared error by up to an order of magnitude over standard techniques, samples gradients (differences of correlated paths) with much lower variance than conventional paths and obtains final images with Poisson reconstruction. A theoretical analysis provides insights to the method.

Markus Kettunen
Aalto University

Marco Manzi
University of Bern

Miika Aittala
Aalto University

Jaakko Lehtinen
Aalto University, Massachusetts Institute of Technology

Frédo Durand
Massachusetts Institute of Technology

Matthias Zwicker
University of Bern

Variance Analysis for Monte Carlo Integration

This paper proposes a new spectral analysis tool that allows calculation of variance in Monte Carlo integration. It estimates the variance convergence rate of various state-of-the-art sampling patterns in both the Euclidean and spherical domains and formulates design principles for conceiving future sampling methods.

Adrien Pilleboue
Université de Lyon 1

Gurprit Singh
Université de Lyon 1

David Coeurjolly
Université de Lyon 1

Michael Kazhdan
Johns Hopkins University

Victor Ostromoukhov
Université de Lyon 1