BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Seoul
X-LIC-LOCATION:Asia/Seoul
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:KST
DTSTART:18871231T000000
DTSTART:19881009T020000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230103T035311Z
LOCATION:Room 325-AB\, Level 3\, West Wing
DTSTART;TZID=Asia/Seoul:20221208T110000
DTEND;TZID=Asia/Seoul:20221208T123000
UID:siggraphasia_SIGGRAPH Asia 2022_sess168_papers_304@linklings.com
SUMMARY:Neural Wavelet-domain Diffusion for 3D Shape Generation
DESCRIPTION:Technical Papers\n\nNeural Wavelet-domain Diffusion for 3D Sha
 pe Generation\n\nHui, Li, Hu, Fu\n\nThis paper presents a new approach for
  3D shape generation, enabling a direct generative modeling on a continuou
 s implicit representation in wavelet frequency domain. Specifically, we pr
 opose a compact wavelet representation with a pair of coarse and detail co
 efficient volumes to implicitly represent 3D shapes via truncated signed d
 istance function and multi-scale biorthogonal wavelet. Then, we formulate 
 a pair of neural networks: a generator based on the diffusion model for pr
 oducing diverse shapes in the form of coarse coefficient volume; and a det
 ail predictor to further produce compatible detail coefficient volumes for
  enriching the generated shapes with fine details. Both quantitative and q
 ualitative experimental results manifest the superiority of our approach i
 n generating diverse and high-quality shapes with complex topology and str
 uctures, clean surfaces, and fine details, exceeding the 3D generation cap
 abilities of the state-of-the-art models.\n\nRegistration Category: FULL A
 CCESS, ON-DEMAND ACCESS\n\nLanguage: ENGLISH\n\nFormat: IN-PERSON, ON-DEMA
 ND
URL:https://sa2022.siggraph.org/en/full-program/?id=papers_304&sess=sess16
 8
END:VEVENT
END:VCALENDAR