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:20250110T023313Z
LOCATION:Hall B5 (1)\, B Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241206T114300
DTEND;TZID=Asia/Tokyo:20241206T115400
UID:siggraphasia_SIGGRAPH Asia 2024_sess142_papers_138@linklings.com
SUMMARY:NU-NeRF: Neural Reconstruction of Nested Transparent Objects with 
 Uncontrolled Capture Environment
DESCRIPTION:Technical Papers\n\nJia-Mu Sun (Insititute of Computing Techno
 logy Chinese Academy of Sciences, KIRI Innovations); Tong Wu (Institute of
  Computing Technology, Chinese Academy of Sciences; University of Chinese 
 Academy of Sciences); Ling-Qi Yan (University of California Santa Barbara)
 ; and Lin Gao (Institute of Computing Technology, Chinese Academy of Scien
 ces; University of Chinese Academy of Sciences)\n\nThe reconstruction of t
 ransparent objects is a challenging problem due to the highly noncontinuou
 s and rapidly changing surface color caused by refraction. Existing method
 s rely on special capture devices, dedicated backgrounds, or ground-truth 
 object masks to provide more priors and reduce the ambiguity of the proble
 m. However, it is hard to apply methods with these special requirements to
  real-life reconstruction tasks, like scenes captured in the wild using mo
 bile devices. Moreover, these methods can only cope with solid and homogen
 eous materials, greatly limiting the scope of the application. To solve th
 e problems above, we propose NU-NeRF to reconstruct nested complex transpa
 rent objects requiring no dedicated capture environment or additional inpu
 t. NU-NeRF is built upon a neural signed distance field formulation and le
 verages neural rendering techniques. It consists of two main stages. In St
 age I, the surface color is separated into reflection and refraction. The 
 reflection is decomposed using physically based material and rendering. Th
 e refraction is modeled using a single MLP given the refraction and view d
 irections, which is a simple yet effective solution of refraction modeling
 . This step produces high-fidelity geometry of the outer surface. In stage
  II, we use explicit ray tracing on the reconstructed outer surface for ac
 curate light transport simulation. The surface reconstruction is executed 
 again inside the outer geometry to obtain any inner surface geometry. In t
 his process, a novel transparent interface formulation is used to cope wit
 h different types of transparent surfaces. Experiments conducted on synthe
 tic scenes and real captured scenes show that NU-NeRF is capable of produc
 ing better reconstruction results than previous methods and achieves accur
 ate nested surface reconstruction while requiring no dedicated capture env
 ironment.\n\nRegistration Category: Full Access, Full Access Supporter\n\n
 Language Format: English Language\n\nSession Chair: Maria Larsson (Univers
 ity of Tokyo)
URL:https://asia.siggraph.org/2024/program/?id=papers_138&sess=sess142
END:VEVENT
END:VCALENDAR