Research map / 3D vision / Face and shape
MeshLLM: Empowering Large Language Models to Progressively Understand and Generate 3D Mesh
Shuangkang Fang, I-Chao Shen, Yufeng Wang, Yi-Hsuan Tsai, Yi Yang, Shuchang Zhou, Wenrui Ding, Takeo Igarashi, Ming-Hsuan Yang
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 |
|---|---|---|---|
| add KNN-based Primitive-Mesh data (ablation) | MeshLLM · ShapeNet chair generation | FID: 49.33 -> 42.39 | Improved |
| add KNN-based Primitive-Mesh data (ablation) | MeshLLM · Objaverse-XL mesh understanding | CIDEr: 0.921 -> 1.753 | Improved |
| add semantic-based Primitive-Mesh data (ablation) | MeshLLM · ShapeNet chair generation | FID: 52.76 -> 42.39 | Improved |
| add semantic-based Primitive-Mesh data (ablation) | MeshLLM · Objaverse-XL mesh understanding | CIDEr: 0.782 -> 1.753 | Improved |
| add vertex-to-face prediction task (ablation) | MeshLLM · ShapeNet chair generation | FID: 48.68 -> 42.39 | Improved |
| add vertex-to-face prediction task (ablation) | MeshLLM · Objaverse-XL mesh understanding | CIDEr: 1.229 -> 1.753 | Improved |
| add mesh assembly task (ablation) | MeshLLM · ShapeNet chair generation | FID: 54.26 -> 42.39 | Improved |
| add mesh assembly task (ablation) | MeshLLM · Objaverse-XL mesh understanding | CIDEr: 0.894 -> 1.753 | Improved |
| train KNN PM first then semantic PM instead of reverse order (ablation) | MeshLLM · ShapeNet chair generation | FID: 45.77 -> 42.39 | Improved |
| train KNN PM first then semantic PM instead of reverse order (ablation) | MeshLLM · Objaverse-XL mesh understanding | CIDEr: 1.303 -> 1.753 | Improved |
| MeshLLM vs LLaMA-Mesh baseline (Primitive-Mesh data and progressive tasks) | LLaMA-3.1-8B-Instruct · ShapeNet chair generation | COV: 19.53 -> 47.33 | Improved |
| MeshLLM vs MeshXL encoder-based generator | MeshLLM · ShapeNet chair generation | FID: 40.33 vs 42.39 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- MeshXL — MeshXL: Neural Coordinate Field for Generative 3D Foundation Models (NeurIPS 2024) · better on 2 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- MeshCoder: LLM-Powered Structured Mesh Code Generation from Point Clouds (NeurIPS 2025 · similarity 0.91)
- 3D-PLOT-LLM: Part-Level Object Tokens for 3D Large Language Models (NeurIPS 2026 · similarity 0.89)
- MeshGPT: Generating Triangle Meshes with Decoder-Only Transformers (CVPR 2024 · similarity 0.89)
- ShapeLLM-Omni: A Native Multimodal LLM for 3D Generation and Understanding (NeurIPS 2025 · similarity 0.87)
- Let Language Constrain Geometry: Vision–Language Models as Semantic and Spatial Critics for 3D Generation (ICML 2026 · similarity 0.87)
- Merge3D: Efficient 3D Multimodal LLMs via Joint 2D-3D Token Merging (CVPR 2026 · similarity 0.87)
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.
