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 B7 (1)\, B Block\, Level 7
DTSTART;TZID=Asia/Tokyo:20241204T172800
DTEND;TZID=Asia/Tokyo:20241204T174000
UID:siggraphasia_SIGGRAPH Asia 2024_sess123_papers_1038@linklings.com
SUMMARY:Multi-Resolution Real-Time Deep Pose-Space Deformation
DESCRIPTION:Technical Papers\n\nMianlun Zheng and Jernej Barbic (Universit
 y of Southern California)\n\nWe present a hard-real-time multi-resolution 
 mesh shape deformation technique for skeleton-driven soft-body characters.
  Producing mesh deformations at multiple levels of detail is very importan
 t in many applications in computer graphics. Our work targets applications
  where the multi-resolution shapes must be generated at fast speeds ("hard
 -real-time", e.g., a few milliseconds at most and preferably under 1 milli
 second), as commonly needed in computer games, virtual reality and Metaver
 se applications. We assume that the character mesh is driven by a skeleton
 , and that high-quality character shapes are available in a set of trainin
 g poses originating from a high-quality (slow) rig such as volumetric FEM 
 simulation. Our method combines multi-resolution analysis, mesh partition 
 of unity, and neural networks, to learn the pre-skinning shape deformation
 s in an arbitrary character pose. Combined with linear blend skinning, thi
 s makes it possible to reconstruct the training shapes, as well as interpo
 late and extrapolate them. Crucially, we simultaneously achieve this at ha
 rd real-time rates and at multiple mesh resolution levels. Our technique m
 akes it possible to trade deformation quality for memory and computation s
 peed, to accommodate the strict requirements of modern real-time systems. 
 Furthermore, we propose memory layout and code improvements to boost compu
 tation speeds. Previous methods for real-time approximations of quality sh
 ape deformations did not focus on hard real-time, or did not investigate t
 he multi-resolution aspect of the problem. Compared to a ``naive'' approac
 h of separately processing each hierarchical level of detail, our method o
 ffers a substantial memory reduction as well as computational speedups. It
  also makes it possible to construct the shape progressively level by leve
 l and interrupt the computation at any time, enabling graceful degradation
  of the deformation detail. We demonstrate our technique on several exampl
 es, including a stylized human character, human hands, and an inverse-kine
 matics-driven quadruped animal.\n\nRegistration Category: Full Access, Ful
 l Access Supporter\n\nLanguage Format: English Language\n\nSession Chair: 
 Huamin Wang (Style3D Research)
URL:https://asia.siggraph.org/2024/program/?id=papers_1038&sess=sess123
END:VEVENT
END:VCALENDAR