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:20250110T023312Z
LOCATION:Hall B5 (2)\, B Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241205T104500
DTEND;TZID=Asia/Tokyo:20241205T105900
UID:siggraphasia_SIGGRAPH Asia 2024_sess128_papers_966@linklings.com
SUMMARY:AdR-Gaussian: Accelerating Gaussian Splatting with Adaptive Radius
DESCRIPTION:Technical Papers\n\nXinzhe Wang, Ran Yi, and Lizhuang Ma (Shan
 ghai Jiao Tong University)\n\n3D Gaussian Splatting (3DGS) is a recent exp
 licit 3D representation that has achieved high-quality reconstruction and 
 real-time rendering of complex scenes. However, the rasterization pipeline
  still suffers from unnecessary overhead resulting from avoidable serial G
 aussian culling, and uneven load due to the distinct number of Gaussian to
  be rendered across pixels, which hinders wider promotion and application 
 of 3DGS. In order to accelerate Gaussian splatting, we propose AdR-Gaussia
 n, which moves part of serial culling in Render stage into the earlier Pre
 process stage to enable parallel culling, employing adaptive radius to nar
 row the rendering pixel range for each Gaussian, and introduces a load bal
 ancing method to minimize thread waiting time during the pixel-parallel re
 ndering. Our contributions are threefold, achieving a rendering speed of 3
 10% while maintaining equivalent or even better quality than the state-of-
 the-art. Firstly, we propose to early cull Gaussian-Tile pairs of low spla
 tting opacity based on an adaptive radius in the Gaussian-parallel Preproc
 ess stage, which reduces the number of affected tile through the Gaussian 
 bounding circle, thus reducing unnecessary overhead and achieving faster r
 endering speed. Secondly, we further propose early culling based on axis-a
 ligned bounding box for Gaussian splatting, which achieves a more signific
 ant reduction in ineffective expenses by accurately calculating the Gaussi
 an size in the 2D directions. Thirdly, we propose a balancing algorithm fo
 r pixel thread load, which compresses the information of heavy-load pixels
  to reduce thread waiting time, and enhance information of light-load pixe
 ls to hedge against rendering quality loss. Experiments on three datasets 
 demonstrate that our algorithm can significantly improve the Gaussian Spla
 tting rendering speed.\n\nRegistration Category: Full Access, Full Access 
 Supporter\n\nLanguage Format: English Language\n\nSession Chair: Manolis S
 avva (Simon Fraser University)
URL:https://asia.siggraph.org/2024/program/?id=papers_966&sess=sess128
END:VEVENT
END:VCALENDAR