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:20251218T030657Z
LOCATION:Meeting Room S426+S427\, Level 4
DTSTART;TZID=Asia/Hong_Kong:20251218T110100
DTEND;TZID=Asia/Hong_Kong:20251218T111200
UID:siggraphasia_SIGGRAPH Asia 2025_sess152_papers_1988@linklings.com
SUMMARY:Neural Texture Splatting: Expressive 3D Gaussian Splatting for Vie
 w Synthesis, Geometry, and Dynamic Reconstruction
DESCRIPTION:Yiming Wang, Shaofei Wang, Marko Mihajlovic, and Siyu Tang (ET
 H Zürich)\n\n3D Gaussian Splatting (3DGS) has emerged as a leading approac
 h for high-quality novel view synthesis, with numerous variants extending 
 its applicability to a broad spectrum of 3D and 4D scene reconstruction ta
 sks. Despite its success, the representational capacity of 3DGS remains li
 mited by the use of 3D Gaussian kernels to model local variations. Recent 
 works have proposed to augment 3DGS with additional per-primitive capacity
 , such as per-splat textures, to enhance its expressiveness. However, thes
 e per-splat texture approaches primarily target dense novel view synthesis
  with a reduced number of Gaussian primitives, and their effectiveness ten
 ds to diminish when applied to more general reconstruction scenarios. In t
 his paper, we aim to achieve concrete performance improvement over state-o
 f-the-art 3DGS variants across a wide range of reconstruction tasks, inclu
 ding novel view synthesis, geometry and dynamic reconstruction, under both
  sparse and dense input settings. To this end, we introduce Neural Texture
  Splatting (NTS). At the core of our approach is a global neural field (re
 presented as a hybrid of a tri-plane and a neural decoder) that predicts l
 ocal appearance and geometric texture fields for each primitive. By levera
 ging this shared global representation that models local texture fields ac
 ross primitives, we significantly reduce model size and facilitate efficie
 nt global information exchange, demonstrating strong generalization across
  tasks. Furthermore, our neural modeling of local texture fields introduce
 s expressive view- and time-dependent effects, a critical aspect that exis
 ting methods fail to account for. Extensive experiments show that Neural T
 exture Splatting consistently improves models and achieves state-of-the-ar
 t results across multiple benchmarks.\n\nRegistration Category: Full Acces
 s, Full Access Supporter\n\nSession Chair: Lin Gao (University of Chinese 
 Academy of Sciences)\n\n
END:VEVENT
END:VCALENDAR