BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Tokyo
X-LIC-LOCATION:Asia/Tokyo
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:JST
DTSTART:18871231T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20250110T023313Z
LOCATION:Hall B5 (2)\, B Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241206T091100
DTEND;TZID=Asia/Tokyo:20241206T092300
UID:siggraphasia_SIGGRAPH Asia 2024_sess140_papers_175@linklings.com
SUMMARY:Differentiable Owen Scrambling
DESCRIPTION:Technical Papers\n\nBastien Doignies (Université Claude Bernar
 d Lyon, LIRIS); David Coeurjolly, Nicolas Bonneel, and Julie Digne (CNRS, 
 LIRIS); and Jean-Claude Iehl and Victor Ostromoukhov (Université Claude Be
 rnard Lyon, LIRIS)\n\nQuasi-Monte Carlo integration is at the core of rend
 ering. This technique estimates the value of an integral by evaluating the
  integrand at well-chosen sample locations. These sample points are design
 ed to cover the domain as uniformly as possible to achieve better converge
 nce rates than purely random points. Deterministic low-discrepancy sequenc
 es have been shown to outperform many competitors by guaranteeing good uni
 formity as measured by the so-called discrepancy metric, and, indirectly, 
 by an integer 𝑡 value relating the number of points falling into each doma
 in stratum with the stratum area (lower 𝑡 is better). To achieve randomnes
 s, scrambling techniques produce multiple realizations preserving the 𝑡 va
 lue, making the construction stochastic. Among them, Owen scrambling is a 
 popular approach that recursively permutes intervals for each dimension. H
 owever, relying on permutation trees makes it incompatible with smooth opt
 imization frameworks. We present a differentiable Owen scrambling that reg
 ularizes permutations. We show that it can effectively be used with automa
 tic differentiation tools for optimizing low-discrepancy sequences to impr
 ove metrics such as optimal transport uniformity, integration error, desig
 ned power spectra or projective properties, while maintaining their initia
 l 𝑡-value as guaranteed by Owen scrambling. In some rendering settings, we
  show that our optimized sequences improve the rendering error.\n\nRegistr
 ation Category: Full Access, Full Access Supporter\n\nLanguage Format: Eng
 lish Language\n\nSession Chair: Seungyong Lee (POSTECH)
URL:https://asia.siggraph.org/2024/program/?id=papers_175&sess=sess140
END:VEVENT
END:VCALENDAR