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:20230103T035307Z
LOCATION:Room 324\, Level 3\, West Wing
DTSTART;TZID=Asia/Seoul:20221206T153000
DTEND;TZID=Asia/Seoul:20221206T170000
UID:siggraphasia_SIGGRAPH Asia 2022_sess155_papers_409@linklings.com
SUMMARY:Learning-based Inverse Rendering of Complex Indoor Scenes with Dif
 ferentiable Monte Carlo Raytracing
DESCRIPTION:Technical Communications, Technical Papers\n\nLearning-based I
 nverse Rendering of Complex Indoor Scenes with Differentiable Monte Carlo 
 Raytracing\n\nZhu, Luan, Huo, Lin, Zhong...\n\nIndoor scenes typically exh
 ibit complex, spatially-varying appearance from global illumination, makin
 g the inverse rendering a challenging ill-posed problem. This work present
 s an end-to-end, learning-based inverse rendering framework incorporating 
 differentiable Monte Carlo raytracing with importance sampling. The framew
 ork takes a single image as input to jointly recover the underlying geomet
 ry, spatially-varying lighting, and photorealistic materials. Specifically
 , we introduce a physically-based differentiable rendering layer with scre
 en-space ray tracing, resulting in more realistic specular reflections tha
 t match the input photo. In addition, we create a large-scale, photorealis
 tic indoor scene dataset with significantly richer details like complex fu
 rniture and dedicated decorations. Further, we design a novel out-of-view 
 lighting network with uncertainty-aware refinement leveraging hypernetwork
 -based neural radiance fields to predict lighting outside the view of the 
 input photo. Through extensive evaluations on common benchmark datasets, w
 e demonstrate superior inverse rendering quality of our method compared to
  state-of-the-art baselines, enabling various applications such as complex
  object insertion and material editing with high fidelity. Code and data w
 ill be made available at https://jingsenzhu.github.io/invrend.\n\nRegistra
 tion Category: FULL ACCESS, ON-DEMAND ACCESS\n\nLanguage: ENGLISH\n\nForma
 t: IN-PERSON, ON-DEMAND
URL:https://sa2022.siggraph.org/en/full-program/?id=papers_409&sess=sess15
 5
END:VEVENT
END:VCALENDAR