TaoFlowForge: Progressive Native Mesh Generation via Cascaded Flow Matching

2026-09-29

AI Summary

This study addresses the challenge of generating production-grade 3D meshes that are simultaneously lightweight, editable, and topologically clean by decomposing mesh generation into vertex generation and connectivity prediction. Methodologically, it proposes a two-stage coarse-to-fine vertex generation strategy combined with connectivity affinity estimation and normal prediction for topology reconstruction. Training employs a cascaded flow matching framework supported by a large-scale hybrid data curation pipeline to facilitate image-conditioned generation. Experimental results demonstrate that the proposed approach outperforms autoregressive baselines in image-conditioned generation and achieves state-of-the-art performance on open-source mesh topology generation tasks. The code and model weights have been made publicly available.

Abstract

3D content generation technology has significantly advanced the work of designers, as well as the 3D printing and gaming industries. However, it remains difficult to produce lightweight, editable, and topologically clean artistic content that is directly production-ready. To achieve this, we present TaoFlowForge, an artistic mesh foundation model that generates production-ready meshes. Specifically, TaoFlowForge decomposes the mesh generation process into vertices generation and their connectivity prediction, i.e., edges. We formulate vertices generation as a two-stage coarse-to-fine process and incorporate several effective loss functions to further enhance its performance. In the connectivity prediction stage, we propose a simple yet effective method for estimating the connectivity affinity between vertices and additionally predict per-vertex normals, which determines the correct orientation of faces. Besides, we construct a large-scale dataset combining hand-crafted 3D assets with public high-quality topology datasets. Based on this, a carefully designed data curation pipeline is employed to filter the raw dataset, retaining only high-quality topology data for model training. Our model is trained on the combined dataset and tested on both out-of-distribution hand-crafted set of 3D assets and public datasets. Under image-conditioned generation, TaoFlowForge outperforms autoregressive methods and achieves state-of-the-art results among open-source mesh topology generators. We will release all the code and weights together with a portion of our test dataset.

Problem

• 3D mesh generation
• production-ready meshes
• topology
• artistic content
• 3D assets

Innovation

• Flow Matching
• Mesh Generation
• Connectivity Prediction
• Data Curation Pipeline
• Coarse-to-Fine