Research map / 3D vision / Face and shape
QuadGPT: Native Quadrilateral Mesh Generation with Autoregressive Models
Jian Liu, Chunshi Wang, Song Guo, Haohan Weng, Zhen Zhou, Zhiqi Li, Jiaao Yu, Yiling Zhu, Jing Xu, Biwen Lei, Zhuo Chen, Chunchao Guo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| initialize from triangle-pretrained model (curriculum) instead of training from scratch (ablation) | QuadGPT · Dense meshes | Chamfer distance: 0.081 -> 0.065 | Improved |
| use truncated DPO with basic fracture penalty instead of standard DPO (ablation) | QuadGPT · Dense meshes | user study score: 1.1 -> 3.3 | Improved |
| add full topological reward (edge loops) to tDPO (ablation) | QuadGPT · Dense meshes | user study score: 3.3 -> 3.7 | Improved |
| apply standard DPO on full meshes after finetuning (ablation) | QuadGPT · Dense meshes | Chamfer distance: 0.065 -> 0.073 | Hurt |
| generate quads natively instead of triangle generation plus conversion (ablation) | TriGPT+RL · Native vs conversion | quad ratio (%): 72% -> 80% | Improved |
| generate quads natively instead of triangles plus conversion (ablation) | TriGPT+RL · Native vs conversion | Chamfer distance: 0.051 -> 0.057 | Hurt |
| use direct coordinate tokens instead of compressed BPT-style tokenizer (ablation) | TriGPT · Triangle meshes | Chamfer distance: 0.078 -> 0.062 | Improved |
| train on only open-source data (QuadGPT-OS) (ablation) | QuadGPT · Dense and artist meshes | quality | Hurt |
| use QuadGPT instead of QuadriFlow | QuadriFlow · Dense meshes | Chamfer distance: 0.045 -> 0.057 | Hurt |
| use QuadGPT instead of BPT | BPT · Artist meshes | Chamfer distance: 0.051 -> 0.043 | Improved |
Closest papers by meaning
- MeshGPT: Generating Triangle Meshes with Decoder-Only Transformers (CVPR 2024 · similarity 0.89)
- ARMesh: Autoregressive Mesh Generation via Next-Level-of-Detail Prediction (NeurIPS 2025 · similarity 0.87)
- QUADify: Extracting Meshes with Pixel-level Details and Materials from Images (CVPR 2024 · similarity 0.86)
- Mesh-RFT: Enhancing Mesh Generation via Fine-grained Reinforcement Fine-Tuning (NeurIPS 2025 · similarity 0.86)
- MeshArt: Generating Articulated Meshes with Structure-Guided Transformers (CVPR 2025 · similarity 0.85)
- CraftsMan3D: High-fidelity Mesh Generation with 3D Native Diffusion and Interactive Geometry Refiner (CVPR 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
