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:20241206T090000
DTEND;TZID=Asia/Tokyo:20241206T091100
UID:siggraphasia_SIGGRAPH Asia 2024_sess140_papers_422@linklings.com
SUMMARY:Differentiating Variance for Variance-Aware Inverse Rendering
DESCRIPTION:Technical Papers\n\nKai Yan (University of California Irvine, 
 Wētā FX); Vincent Pegoraro, Marc Droske, and Jiří Vorba (Wētā FX); and Shu
 ang Zhao (University of California Irvine)\n\nMonte Carlo methods have bee
 n widely adopted in physics-based rendering.\n    A key property of a Mont
 e Carlo estimator is its variance, which dictates the convergence rate of 
 the estimator.\n    In this paper, we devise a mathematical formulation fo
 r derivatives of rendering variance with respect to not only scene paramet
 ers (e.g., surface roughness) but also sampling probabilities.\n    Based 
 on this formulation, we introduce unbiased Monte Carlo estimators for thos
 e derivatives.\n    Our theory and algorithm enable variance-aware inverse
  rendering which alters a virtual scene and/or an estimator in an optimal 
 way to offer a good balance between bias and variance.\n    We evaluate ou
 r technique using several synthetic examples.\n\nRegistration Category: Fu
 ll Access, Full Access Supporter\n\nLanguage Format: English Language\n\nS
 ession Chair: Seungyong Lee (POSTECH)
URL:https://asia.siggraph.org/2024/program/?id=papers_422&sess=sess140
END:VEVENT
END:VCALENDAR