TopGen: Learning Structural Layouts and Cross-Fields for Quadrilateral Mesh Generation

TL;DR

TopGen is proposed, a robust and efficient learning-based framework that mimics professional manual modeling workflows by simultaneously predicting structural layouts and cross-fields and significantly outperforms existing state-of-the-art methods in both geometric fidelity and topological edge flow rationality.

Abstract

High-quality quadrilateral mesh generation is a fundamental challenge in computer graphics. Traditional optimization-based methods are often constrained by the topological quality of input meshes and suffer from severe efficiency bottlenecks, frequently becoming computationally prohibitive when handling high-resolution models. While emerging learning-based approaches offer greater flexibility, they primarily focus on cross-field prediction, often resulting in the loss of critical structural layouts and a lack of editability. In this paper, we propose TopGen, a robust and efficient learning-based framework that mimics professional manual modeling workflows by simultaneously predicting structural layouts and cross-fields. By processing input triangular meshes through point cloud sampling and a shape encoder, TopGen is inherently robust to non-manifold geometries and low-quality initial topologies. We introduce a dual-query decoder using edge-based and face-based sampling points as queries to perform structural line classification and cross-field regression in parallel. This integrated approach explicitly extracts the geometric skeleton while concurrently capturing orientation fields. Such synergy ensures the preservation of geometric integrity and provides an intuitive, editable foundation for subsequent quadrilateral remeshing. To support this framework, we also introduce a large-scale quadrilateral mesh dataset, TopGen-220K, featuring high-quality paired data comprising raw triangular meshes, structural layouts, cross-fields, and their corresponding quad meshes. Experimental results demonstrate that TopGen significantly outperforms existing state-of-the-art methods in both geometric fidelity and topological edge flow rationality.

Figures (9)

• Figure 1: TopGen achieves high-quality quadrilateral remeshing across diverse geometries, ranging from organic to mechanical models and AI-generated meshes.
• Figure 2: A sample from TopGen-220K dataset. From left to right: the input triangular mesh, the structural layout, the cross-field, and the quadrilateral mesh.
• Figure 3: Distribution of TopGen-220K.
• Figure 4: Overview of the TopGen pipeline. Our framework first samples point clouds from the input mesh via the SES strategy chen2025dora. A geometry-aware encoder then maps these points into a latent space. Subsequently, our proposed Dual-Query Decoder concurrently decodes the structural layouts and cross-fields in parallel. Finally, both predictions serve as joint guidance to facilitate high-quality quadrilateral remeshing.
• Figure 5: Topology-aware neighborhoods.