postPerspective

Roll the Dice

Tuesday, 26 July, 9:00 am - 10:30 am, Anaheim Convention Center, Room 303 A-C

Blue-Noise Dithered Sampling

A method that improves the visual fidelity of Monte Carlo renderings without increasing the sampling effort. By correlating the samples between pixels using specially constructed blue-noise matrices, the method minimizes the low-frequency content in the distribution of the approximation error, thereby reducing the perceptual error of the image.

Iliyan Georgiev
Solid Angle S.L.

Marcos Fajardo
Solid Angle S.L.

Cache-Friendly Micro-Jittered Sampling

Introducing a cache-friendly micro-jittered technique for faster multi-dimensional Monte Carlo integration in parallel ray tracing engines. The method is simple and compatible with any low-discrepancy sequences, and it can drastically reduce rendering times of GPU (even CPU) path tracers or any stochastic ray-based renderers.

Arthur Dufay
Technicolor, LP2N (CNRS)

Pascal Lecocq
Technicolor

Jean-Eudes Marvie
Technicolor

Romain Pacanowski
LP2N (CNRS)

Xavier Granier
LP2N (CNRS)

Stochastic Layered Alpha Blending

By exploring the continuum of order-independent transparency techniques, NVIDIA identified a new algorithm that selects benefits from both stochastic transparency and k-buffering. Stochastic layered alpha blending provides faster consistent and unbiased convergence, and an explicit parameter that reduces noise in exchange for adding bias.

Chris Wyman
NVIDIA Corporation

Quantum Supersampling

With a focus on practical use of near-future quantum logic devices in computer graphics, this project implements a qubit-based supersampling technique with advantages over Monte Carlo estimation, designed for physical fabrication as an integrated-silicon photonic device.

Eric Johnston
University of Bristol, Solid Angle