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:20260817T171534Z
LOCATION:Hall B5 (2)\, B Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241203T163000
DTEND;TZID=Asia/Tokyo:20241203T164100
UID:siggraphasia_SIGGRAPH Asia 2024_sess110_papers_669@linklings.com
SUMMARY:MARS: Multi-sample Allocation through Russian roulette and Splitti
 ng
DESCRIPTION:Joshua Meyer, Alexander Rath, and Ömercan Yazici (Saarland Inf
 ormatics Campus) and Philipp Slusallek (German Research Center for Artific
 ial Intelligence, Saarland Informatics Campus)\n\nMultiple importance samp
 ling (MIS) is an indispensable tool in rendering that constructs robust sa
 mpling strategies by combining the respective strengths of individual dist
 ributions. Its efficiency can be greatly improved by carefully selecting t
 he number of samples drawn from each distribution, but automating this pro
 cess remains a challenging problem. Existing works are mostly limited to m
 ixture sampling, in which only a single sample is drawn in total, and the 
 works that do investigate multi-sample MIS only optimize the sample counts
  at a per-pixel level, which cannot account for variations beyond the firs
 t bounce. Recent work on Russian roulette and splitting has demonstrated h
 ow fixed-point schemes can be used to spatially vary sample counts to opti
 mize image efficiency but is limited to choosing the same number of sample
 s across all sampling strategies. Our work proposes a highly flexible samp
 le allocation strategy that bridges the gap between these areas of work. W
 e show how to iteratively optimize the sample counts to maximize the effic
 iency of the rendered image using a lightweight data structure, which allo
 ws us to make local and individual decisions per technique. We demonstrate
  the benefits of our approach in two applications, path guiding and bidire
 ctional path tracing, in both of which we achieve consistent and substanti
 al speedups over the respective previous state-of-the-art.\n\nRegistration
  Category: Full Access, Full Access Supporter\n\nLanguage Format: English 
 Language\n\nSession Chair: Michael Wimmer (TU Wien, Technische Universität
  Wien (TU Wien))\n\n
URL:https://asia.siggraph.org/2024/program/?id=papers_669&sess=sess110
END:VEVENT
END:VCALENDAR