BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Seoul
X-LIC-LOCATION:Asia/Seoul
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:KST
DTSTART:18871231T000000
DTSTART:19881009T020000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230103T035312Z
LOCATION:Room 324\, Level 3\, West Wing
DTSTART;TZID=Asia/Seoul:20221209T103000
DTEND;TZID=Asia/Seoul:20221209T120000
UID:siggraphasia_SIGGRAPH Asia 2022_sess173_papers_520@linklings.com
SUMMARY:Deep Adaptive Sampling and Reconstruction using Analytic Distribut
 ions
DESCRIPTION:Technical Communications, Technical Papers\n\nDeep Adaptive Sa
 mpling and Reconstruction using Analytic Distributions\n\nSalehi, Manzi, R
 oethlin, Schroers, Weber...\n\nWe propose an adaptive sampling and reconst
 ruction method for offline Monte Carlo rendering. Our method produces samp
 ling maps constrained by a user-defined budget that minimize the expected 
 future denoising error. Compared to other state-of-the-art methods, which 
 produce the necessary training data on the fly by composing pre-rendered i
 mages, our method samples from analytic noise distributions instead. These
  distributions are compact and closely approximate the pixel value distrib
 utions stemming from Monte Carlo rendering. Our method can efficiently sam
 ple training data by leveraging only a few per-pixel statistics of the tar
 get distribution, which provides several benefits over the current state o
 f the art. Most notably, our analytic distributions’ modeling accuracy and
  sampling efficiency increase with sample count, essential for high-qualit
 y offline rendering. Although our distributions are approximate, our metho
 d supports joint end-to-end training of the sampling and denoising network
 s. Finally, we propose the addition of a global summary module to our arch
 itecture that accumulates valuable information from image regions outside 
 of the network’s receptive field. This information discourages sub-optimal
  decisions based on local information. Our evaluation against other state-
 of-the-art neural sampling methods demonstrates denoising quality and data
  efficiency improvements.\n\nRegistration Category: FULL ACCESS, ON-DEMAND
  ACCESS\n\nLanguage: ENGLISH\n\nFormat: IN-PERSON, ON-DEMAND
URL:https://sa2022.siggraph.org/en/full-program/?id=papers_520&sess=sess17
 3
END:VEVENT
END:VCALENDAR