MeshTailor: Cutting Seams via Generative Mesh Traversal

TL;DR

MeshTailor is a novel mesh-native generative framework that synthesizes edge-aligned seams on 3D surfaces, improving coherence and structural regularity over prior methods.

Contribution

It introduces a hierarchical seam graph serialization and a dual-stream transformer for direct mesh-based seam generation, avoiding projection artifacts.

Findings

Produces more coherent seam layouts than recent baselines.

Operates directly on mesh graphs, eliminating projection artifacts.

Uses a hierarchical representation and dual-stream embeddings for better results.

Abstract

We present MeshTailor, the first mesh-native generative framework for synthesizing edge-aligned seams on 3D surfaces. Unlike prior optimization-based or extrinsic learning-based methods, MeshTailor operates directly on the mesh graph, eliminating projection artifacts and fragile snapping heuristics. We introduce ChainingSeams, a hierarchical serialization of the seam graph that orders chains from global structural cuts down to local details in a coarse-to-fine manner, and a dual-stream encoder that fuses topological and geometric context. Leveraging this hierarchical representation and dual-stream vertex embeddings, our MeshTailor Transformer utilizes an autoregressive pointer layer to trace seams vertex-by-vertex within local neighborhoods. Extensive evaluations show that MeshTailor produces more coherent and structurally regular seam layouts compared to recent optimization-based and learning-based baselines.