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:20251218T030655Z
LOCATION:Meeting Room S426+S427\, Level 4
DTSTART;TZID=Asia/Hong_Kong:20251218T135300
DTEND;TZID=Asia/Hong_Kong:20251218T140400
UID:siggraphasia_SIGGRAPH Asia 2025_sess156_papers_1217@linklings.com
SUMMARY:A compact stochastic representation for Monte Carlo Path Traced im
 ages
DESCRIPTION:Matthias Sebastian Treder, Pavlos Makridis, Alexis Lechat, Jes
 us Zarzar, Marina Villanueva Barreiro, and Roc Ramon Currius (Sony Interac
 tive Entertainment)\n\nWe present a compact, learning-based representation
  that captures the full Monte Carlo sampling distribution of a rendered im
 age. Our approach enables rendering at arbitrary samples per pixel (SPP) d
 uring inference without requiring expensive path tracing operations. This 
 is achieved by fitting parametric distributions to per-pixel radiance valu
 es, which can be efficiently estimated, stored, and sampled.\nOur method p
 roceeds in three stages. First, we map radiance samples into radial log sp
 ace, which encourages Gaussian-like distributions while preserving angular
  relationships. Second, we fit each pixel’s distribution using 3D Gaussian
  Mixture Models (GMMs), trained online with minimal memory overhead, makin
 g the approach compatible with standard path tracers. For inference, we in
 troduce an optimized sampling scheme whose complexity is independent of th
 e target SPP, enabling fast synthesis of high-SPP images. Additionally, we
  demonstrate that the learned representations can be heavily compressed us
 ing quantization and codebook techniques with negligible quality loss. Exp
 eriments show that GMMs strike an effective balance between expressiveness
  and sparsity. Compared to alternative models, our method better captures 
 pixel-wise Monte Carlo distributions. Lastly, we illustrate the versatilit
 y of our representation with applications such as firefly rejection and ra
 y-distribution-driven denoising.\n\nRegistration Category: Full Access, Fu
 ll Access Supporter\n\nSession Chair: Manyi Li (Shandong University)\n\n
END:VEVENT
END:VCALENDAR