High Resolution UDF Meshing via Iterative Networks

2025-09-21

AI Summary

High-resolution unsigned distance field (UDF) meshing suffers from strong noise interference and local voxel modeling that neglects neighborhood information, leading to surface incompleteness and holes. Method: We propose the first iterative neural meshing framework, which dynamically aggregates spatial neighborhood information—including surface locations, distance values, and gradients—across multiple inference rounds. Each iteration refines erroneous classifications and enhances stability in ambiguous or geometrically complex regions. Contribution/Results: Our core innovation integrates implicit field sampling, gradient-guided surface detection, and cross-iteration contextual propagation, enabling multi-level context-aware voxel-wise surface recovery. Evaluated on diverse high-detail, topologically complex 3D models, our method significantly outperforms existing single-pass extraction approaches: reconstructed meshes exhibit superior completeness and geometric fidelity, especially in fine-scale structures and thin-wall regions.

Technology Category

• Machine Learning: Learning with Manifolds
• Computer Vision: 3D Computer Vision
• Natural Language Processing: Information Extraction

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Unsigned Distance Fields (UDFs) are a natural implicit representation for open surfaces but, unlike Signed Distance Fields (SDFs), are challenging to triangulate into explicit meshes. This is especially true at high resolutions where neural UDFs exhibit higher noise levels, which makes it hard to capture fine details. Most current techniques perform within single voxels without reference to their neighborhood, resulting in missing surface and holes where the UDF is ambiguous or noisy. We show that this can be remedied by performing several passes and by reasoning on previously extracted surface elements to incorporate neighborhood information. Our key contribution is an iterative neural network that does this and progressively improves surface recovery within each voxel by spatially propagating information from increasingly distant neighbors. Unlike single-pass methods, our approach integrates newly detected surfaces, distance values, and gradients across multiple iterations, effectively correcting errors and stabilizing extraction in challenging regions. Experiments on diverse 3D models demonstrate that our method produces significantly more accurate and complete meshes than existing approaches, particularly for complex geometries, enabling UDF surface extraction at higher resolutions where traditional methods fail.

Problem

• Triangulating noisy neural UDFs into high-resolution meshes
• Addressing missing surfaces and holes from ambiguous UDF data
• Improving surface recovery by integrating neighborhood information iteratively

Innovation

• Iterative neural network propagates spatial neighbor information
• Multi-pass integration of surfaces, distances and gradients
• Progressive error correction enables high-resolution UDF meshing