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 (1)\, B Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241205T130000
DTEND;TZID=Asia/Tokyo:20241205T131100
UID:siggraphasia_SIGGRAPH Asia 2024_sess130_papers_635@linklings.com
SUMMARY:SpaceMesh: A Continuous Representation for Learning Manifold Surfa
 ce Meshes
DESCRIPTION:Technical Papers\n\nTianchang Shen (University of Toronto, NVI
 DIA Research) and Zhaoshuo Li, Marc Law, Matan Atzmon, Sanja Fidler, James
  Lucas, Jun Gao, and Nicholas Sharp (NVIDIA Research)\n\nMeshes are ubiqui
 tous in visual computing and simulation, yet most existing machine learnin
 g techniques represent meshes only indirectly, e.g. as the level set of a 
 scalar field, or deformation of a template, or as a disordered triangle so
 up lacking local structure. This work presents a scheme to directly genera
 te manifold, polygonal meshes of arbitrary connectivity as the output of a
  neural network. Our key innovation is to define a continuous latent conne
 ctivity space at each mesh vertex, which implies the discrete mesh. In par
 ticular, our vertex embeddings generate cyclic neighbor relationships in a
  halfedge mesh representation, which gives a guarantee of edge-manifoldnes
 s and the ability to represent general polygonal meshes. This representati
 on is well-suited to machine learning and stochastic optimization, without
  restriction on connectivity or topology. We first explore the basic prope
 rties of this representation, then use it to fit distributions of meshes f
 rom large datasets. The resulting models generate diverse meshes with tess
 ellation structure learned from the dataset population, with concise detai
 ls and high-quality mesh elements. In applications, this approach not only
  yields high-quality outputs from generative models, but also enables dire
 ctly learning challenging geometry processing tasks such as mesh repair.\n
 \nRegistration Category: Full Access, Full Access Supporter\n\nLanguage Fo
 rmat: English Language\n\nSession Chair: Noam Aigerman (University of Mont
 real)
URL:https://asia.siggraph.org/2024/program/?id=papers_635&sess=sess130
END:VEVENT
END:VCALENDAR