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:20241205T104500
DTEND;TZID=Asia/Tokyo:20241205T105600
UID:siggraphasia_SIGGRAPH Asia 2024_sess129_papers_1108@linklings.com
SUMMARY:Millimetric Human Surface Capture in Minutes
DESCRIPTION:Technical Papers\n\nBriac Toussaint and Laurence Boissieux (Ce
 ntre Inria de l’Université Grenoble Alpes); Diego Thomas (Kyushu Universit
 y); Edmond Boyer (Meta Reality Labs Research); and Jean-Sébastien Franco (
 LJK, CNRS, Grenoble INP, Université Grenoble Alpes; Centre Inria de l’Univ
 ersité Grenoble Alpes)\n\nDetailed human surface capture from multiple ima
 ges is an essential component for many 3D production, analysis and transmi
 ssion tasks. Yet producing millimetric precision 3D models in practical ti
 me, and actually verifying their 3D accuracy in a real-world capture conte
 xt, remain key challenges due to the lack of specific methods and data for
  these goals. We propose two complementary contributions to this end. The 
 first one is a highly scalable neural surface radiance field approach able
  to achieve millimetric precision by construction, while demonstrating hig
 h compute and memory efficiency. The second one is a novel dataset, MVMann
 equin, of clothed mannequin geometry captured with a high resolution hand-
 held 3D scanner paired with calibrated multi-view images, that allows to v
 erify the millimetric accuracy claim. Although our approach can produce su
 ch highly dense and precise geometry, we show how aggressive sparsificatio
 n and optimizations of the neural surface pipeline allow estimations in mi
 nutes of computation time using only a few GB of GPU memory, while allowin
 g for real-time millisecond neural rendering. On the basis of our framewor
 k and dataset, we show that our method achieves submillimetric accuracy an
 d completeness for 77% of the points in less than three minutes of trainin
 g time, with 68 viewpoints.\n\nRegistration Category: Full Access, Full Ac
 cess Supporter\n\nLanguage Format: English Language\n\nSession Chair: Yuti
 ng Ye (Reality Labs Research, Meta; Meta)
URL:https://asia.siggraph.org/2024/program/?id=papers_1108&sess=sess129
END:VEVENT
END:VCALENDAR