Gamasutra

Efficient Sampling & Rendering

Monday, 25 July, 3:45 pm - 5:35 pm, Anaheim Convention Center, Ballroom C
Session Chair: Peter-Pike Sloan, Activision, Inc.

Integration with Stochastic Point Processes

We present a novel comprehensive approach for studying error in integral estimation based on point process statistics. We derive exact formulae for bias and variance of integral estimates.

Cengiz Oztireli
ETHZ, ETH

Fast 4D Sheared Filtering for Interactive Rendering of Distribution Effects

A fast sheared filtering approach on the GPU for distribution effects, such as depth of field, soft shadows, and diffuse global illumination, reducing sampling rate by 5-8x and increasing performance by 4x.

Ling-Qi Yan
University of California, Berkeley

Soham Uday Mehta
University of California, Berkeley

Ravi Ramamoorthi
University of California, Berkeley

Frédo Durand
Massachusetts Institute of Technology

Adaptive Polynomial Rendering

A new adaptive rendering method to improve the performance of Monte Carlo ray tracing, by reducing noise in rendered images while preserving high-frequency edges. The method approximates an image with polynomial functions and minimizes reconstruction error by robustly estimating an optimal order of each polynomial function.

Bochang Moon
The Walt Disney Company

Steven McDonagh
The Walt Disney Company

Kenny Mitchell
The Walt Disney Company

Markus Gross
The Walt Disney Company

Real-Time Polygonal-Light Shading With Linearly Transformed Cosines

This novel technique for real-time polygonal-light shading builds upon the Linearly Transformed Cosine, a new spherical distribution that has a closed-form integral over arbitrary spherical polygons and covers a wide class of spherical shapes. The technique is robust, fast, accurate, and simple to implement.

Eric Heitz
Unity Technologies

Jonathan Dupuy
Unity Technologies

Stephen Hill
Ubisoft Entertainment S.A

David Neubelt
Ready at Dawn Studios

Adjoint-Driven Russian Roulette and Splitting in Light-Transport Simulation

This paper introduces a Russian roulette and splitting framework for path sampling in light-transport simulations that is based on a well-founded theory and shows its remarkable importance-sampling capability when used by itself as well as in synergistic combination with importance-driven directional sampling methods.

Jiri Vorba
Weta Digital Ltd, Charles University in Prague

Jaroslav Křivánek
Charles University in Prague