Spelunking the Deep: Guaranteed Queries on General Neural Implicit Surfaces via Range Analysis

  • 2022-06-24 17:45:26
  • Nicholas Sharp, Alec Jacobson
  • 0

Abstract

Neural implicit representations, which encode a surface as the level set of aneural network applied to spatial coordinates, have proven to be remarkablyeffective for optimizing, compressing, and generating 3D geometry. Althoughthese representations are easy to fit, it is not clear how to best evaluategeometric queries on the shape, such as intersecting against a ray or finding aclosest point. The predominant approach is to encourage the network to have asigned distance property. However, this property typically holds onlyapproximately, leading to robustness issues, and holds only at the conclusionof training, inhibiting the use of queries in loss functions. Instead, thiswork presents a new approach to perform queries directly on general neuralimplicit functions for a wide range of existing architectures. Our key tool isthe application of range analysis to neural networks, using automaticarithmetic rules to bound the output of a network over a region; we conduct astudy of range analysis on neural networks, and identify variants of affinearithmetic which are highly effective. We use the resulting bounds to developgeometric queries including ray casting, intersection testing, constructingspatial hierarchies, fast mesh extraction, closest-point evaluation, evaluatingbulk properties, and more. Our queries can be efficiently evaluated on GPUs,and offer concrete accuracy guarantees even on randomly-initialized networks,enabling their use in training objectives and beyond. We also show apreliminary application to inverse rendering.

 

Quick Read (beta)

loading the full paper ...