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:20260817T171533Z
LOCATION:Hall B5 (1)\, B Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241205T165300
DTEND;TZID=Asia/Tokyo:20241205T170500
UID:siggraphasia_SIGGRAPH Asia 2024_sess136_tog_103@linklings.com
SUMMARY:TriHuman: A Real-time and Controllable Tri-plane Representation fo
 r Detailed Human Geometry and Appearance Synthesis
DESCRIPTION:Heming Zhu (Max Planck Institute for Informatics, Saarland Inf
 ormatics Campus); Fangneng Zhan (Max Planck Institute for Informatics); an
 d Christian Theobalt and Marc Habermann (Max Planck Institute for Informat
 ics; Saarbrücken Research Center for Visual Computing, Interaction and AI)
 \n\nCreating controllable, photorealistic, and geometrically detailed digi
 tal doubles of real humans solely from video data is a key challenge in Co
 mputer Graphics and Vision, especially when real-time performance is requi
 red. Recent methods attach a neural radiance field (NeRF) to an articulate
 d structure, e.g., a body model or a skeleton, to map points into a pose c
 anonical space while conditioning the NeRF on the skeletal pose. These app
 roaches typically parameterize the neural field with a multi-layer percept
 ron (MLP) leading to a slow runtime. To address this drawback, we propose 
 TriHuman a novel human-tailored, deformable, and efficient tri-plane repre
 sentation, which achieves real-time performance, state-of-the-art pose-con
 trollable geometry synthesis as well as photorealistic rendering quality. 
 At the core, we non-rigidly warp global ray samples into our undeformed tr
 i-plane texture space, which effectively addresses the problem of global p
 oints being mapped to the same tri-plane locations. We then show how such 
 a tri-plane feature representation can be conditioned on the skeletal moti
 on to account for dynamic appearance and geometry changes. Our results dem
 onstrate a clear step towards higher quality in terms of geometry and appe
 arance modeling of humans and runtime performance.\n\nRegistration Categor
 y: Full Access, Full Access Supporter\n\nLanguage Format: English Language
 \n\nSession Chair: Manolis Savva (Simon Fraser University)\n\n
URL:https://asia.siggraph.org/2024/program/?id=tog_103&sess=sess136
END:VEVENT
END:VCALENDAR