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:20251218T030656Z
LOCATION:Meeting Room S221\, Level 2
DTSTART;TZID=Asia/Hong_Kong:20251215T151100
DTEND;TZID=Asia/Hong_Kong:20251215T152200
UID:siggraphasia_SIGGRAPH Asia 2025_sess111_papers_1755@linklings.com
SUMMARY:NeuVAS: Neural Implicit Surfaces for Variational Shape Modeling
DESCRIPTION:Pengfei Wang (Shandong University); Qiujie Dong (University of
  Hong Kong); Fangtian Liang (Shandong University); Hao Pan (Tsinghua Unive
 rsity); Lei Yang (University of Hong Kong); Congyi Zhang (University of Br
 itish Columbia); Guying Lin (University of Hong Kong); Caiming Zhang, Yuan
 feng Zhou, Changhe Tu, and Shiqing Xin (Shandong University); Alla Sheffer
  (University of British Columbia); and Xin Li and Wenping Wang (Texas A&M 
 University)\n\nNeural implicit shape representation has drawn significant 
 attention in recent years due to its continuity, differentiability, and to
 pological flexibility. However, directly modeling the shape of neural impl
 icit fields, especially the neural signed distance function (SDF), with sp
 arse geometric control is still a challenging task. While 3D curve network
 s can provide intuitive control over explicit surfaces, the sparsity and v
 aried topology of these networks introduce ambiguity in surface shape inte
 rpolation and present challenges in mesh layout design under curve constra
 ints. Consequently, achieving reasonable surfacing from curve networks has
  long been a challenge in mesh modeling. In this paper, we propose NeuVAS,
  a curvature-based approach\nto solve the neural SDF under curve network c
 onstraints. Leveraging the differentiability of neural shape representatio
 ns, we introduce a smooth term to regularize the zero-level surface of the
  SDF, providing dense control over shape interpolation. Typically, a reaso
 nable surface interpolated from a curve network consists of piecewise smoo
 th surface patches that are 𝐶0 continuous at curve constraints. However, e
 ncoding such a shape using a neural SDF poses significant challenges. To c
 onstruct piecewise smoothness on neural SDFs, we minimize an optional smoo
 th term based on curvature in the space between the curves while relaxing 
 this constraint near feature curves. Moreover, our method can accommodate 
 either structured curve\nnetworks or oriented point clouds as input constr
 aints, making it applicable to a broad range of scenarios. A comprehensive
  comparison with existing state-of-the-art methods demonstrates the signif
 icant advantages of our approach in surfacing curve networks.\n\nRegistrat
 ion Category: Full Access, Full Access Supporter\n\nSession Chair: Ligang 
 Liu (University of Science and Technology of China)\n\n
END:VEVENT
END:VCALENDAR