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:20260817T171534Z
LOCATION:Hall B5 (1)\, B Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241206T093400
DTEND;TZID=Asia/Tokyo:20241206T094600
UID:siggraphasia_SIGGRAPH Asia 2024_sess139_papers_393@linklings.com
SUMMARY:Towards Unified 3D Hair Reconstruction from Single-View Portraits
DESCRIPTION:Yujian Zheng, Yuda Qiu, and Leyang Jin (Chinese University of 
 Hong Kong, Shenzhen); Chongyang Ma, Haibin Huang, Di Zhang, and Pengfei Wa
 n (Kuaishou Technology); and Xiaoguang Han (Chinese University of Hong Kon
 g, Shenzhen)\n\nSingle-view 3D hair reconstruction is challenging, due to 
 the wide range of shape variations among diverse hairstyles. Current state
 -of-the-art methods are specialized in recovering un-braided 3D hairs and 
 often take braided styles as their failure cases, because of the inherent 
 difficulty to define priors for complex hairstyles, whether rule-based or 
 data-based. We propose a novel strategy to enable single-view 3D reconstru
 ction for a variety of hair types via a unified pipeline. To achieve this,
  we first collect a large-scale synthetic multi-view hair dataset SynMvHai
 r with diverse 3D hair in both braided and un-braided styles, and learn tw
 o diffusion priors specialized on hair. Then we optimize 3D Gaussian-based
  hair from the priors with two specially designed modules, i.e. view-wise 
 and pixel-wise Gaussian refinement. Our experiments demonstrate that recon
 structing braided and un-braided 3D hair from single-view images via a uni
 fied approach is possible and our method achieves the state-of-the-art per
 formance in recovering complex hairstyles. It is worth to mention that our
  method shows good generalization ability to real images, although it lear
 ns hair priors from synthetic data. Code and data are available at https:/
 /unihair24.github.io\n\nRegistration Category: Full Access, Full Access Su
 pporter\n\nLanguage Format: English Language\n\nSession Chair: Kui Wu (LIG
 HTSPEED)\n\n
URL:https://asia.siggraph.org/2024/program/?id=papers_393&sess=sess139
END:VEVENT
END:VCALENDAR