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:20260817T171532Z
LOCATION:Hall B5 (2)\, B Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241205T133400
DTEND;TZID=Asia/Tokyo:20241205T134600
UID:siggraphasia_SIGGRAPH Asia 2024_sess131_papers_195@linklings.com
SUMMARY:Neural Product Importance Sampling via Warp Composition
DESCRIPTION:Joey Litalien (McGill University) and Miloš Hašan, Fujun Luan,
  Krishna Mullia, and Iliyan Georgiev (Adobe Research)\n\nAchieving high ef
 ficiency in modern photorealistic rendering methods hinges on using Monte 
 Carlo sampling distributions that closely approximate the illumination int
 egral estimated for every pixel. Samples are typically generated from a se
 t of simple distributions, each targeting a different factor in the integr
 and, which are combined via multiple importance sampling. The resulting mi
 xture distribution can be far from the actual product of all factors, lead
 ing to sub-optimal variance even for direct-illumination estimation. We pr
 esent a learning-based method to efficiently importance sample illuminatio
 n product integrals (e.g., the product of environment lighting and materia
 l terms) using normalizing flows. Our neural product sampler composes a fl
 ow head warp with an emitter tail warp. The small conditional head is repr
 esented by a neural spline flow, while the large unconditional tail is dis
 cretized per environment map and its evaluation is instant. If the conditi
 oning is low-dimensional, the head warp can be discretized for even better
  performance. We demonstrate variance reduction over prior methods on a ra
 nge of applications comprising complex geometry, materials and illuminatio
 n.\n\nRegistration Category: Full Access, Full Access Supporter\n\nLanguag
 e Format: English Language\n\nSession Chair: Seungyong Lee (POSTECH)\n\n
URL:https://asia.siggraph.org/2024/program/?id=papers_195&sess=sess131
END:VEVENT
END:VCALENDAR