Presentation

<
SAND: Spatially Adaptive Network Depth for Fast Sampling of Neural Implicit Surfaces
SessionCreating & Iso-Surfacing SDFs
DescriptionSAND is an efficient neural implicit surface representation framework that accelerates sampling by adapting network depth to spatial importance and local geometric complexity. By combining a volumetric depth map with a tailed MLP for adaptive termination, it significantly improves inference-time query efficiency while preserving high-fidelity reconstruction and enabling neural level-of-detail.
Authors
Chuanxiang Yang
Shandong University
Junhui Hou
City University of Hong Kong
Yuan Liu
Hong Kong University of Science and Technology
Siyu Ren
City University of Hong Kong
Guangshun Wei
Shandong University
Taku Komura
University of Hong Kong
Yuanfeng Zhou
Shandong University
Wenping Wang
Texas A&M University
Event Type
Technical Paper
TimeMonday, 20 July 20264:05pm - 4:15pm PDT
LocationRoom 408 B
Digital Library PDF
Session Time & Location
Sunday, 19 July 20266:00pm - 8:45pm PDTHall K
Monday, 20 July 20263:45pm - 5:35pm PDTRoom 408 B
Keywords
Artificial Intelligence/Machine Learning
Geometry
Modeling
Registration Categories
Full Conference Supporter
Full Conference
Next PresentationNext PresentationThe PhaseTree: Multiphase Signed Distance Fields
Similar Presentations
Explicit flows for implicit surfaces
NeuBase: Spline Surfaces with Neural Basis Functions
Uncertainty-aware geometry processing on Gaussian Process Implicit Surfaces