BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Hong_Kong
X-LIC-LOCATION:Asia/Hong_Kong
BEGIN:STANDARD
TZOFFSETFROM:+0800
TZOFFSETTO:+0800
TZNAME:HKT
DTSTART:19911015T033000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20251218T030653Z
LOCATION:Meeting Room S221\, Level 2
DTSTART;TZID=Asia/Hong_Kong:20251215T154400
DTEND;TZID=Asia/Hong_Kong:20251215T155500
UID:siggraphasia_SIGGRAPH Asia 2025_sess111_papers_1744@linklings.com
SUMMARY:Practical Gaussian Process Implicit Surfaces with Sparse Convoluti
 ons
DESCRIPTION:Kehan Xu (Dartmouth College), Benedikt Bitterli and Eugene d'E
 on (NVIDIA), and Wojciech Jarosz (Dartmouth College)\n\nA fundamental chal
 lenge in rendering has been the dichotomy between surface and volume model
 s. Gaussian Process Implicit Surfaces (GPISes) recently provided a unified
  approach for surfaces, volumes, and the spectrum in between. However, thi
 s representation remains impractical due to its high computational cost an
 d mathematical complexity. We address these limitations by reformulating G
 PISes as procedural noise, eliminating expensive linear system solves whil
 e maintaining control over spatial correlations. Our method enables effici
 ent sampling of stochastic realizations and supports flexible conditioning
  of values and derivatives through pathwise updates. To further enable pra
 ctical rendering, we derive analytic distributions for surface normals, al
 lowing for variance-reduced light transport via next-event estimation and 
 multiple importance sampling. Our framework achieves efficient, high-quali
 ty rendering of stochastic surfaces and volumes with significantly simplif
 ied implementations on both CPU and GPU, while preserving the generality o
 f the original GPIS representation.\n\nRegistration Category: Full Access,
  Full Access Supporter\n\nSession Chair: Ligang Liu (University of Science
  and Technology of China)\n\n
END:VEVENT
END:VCALENDAR