Research map / 3D vision / Face and shape
Textured mesh generation
Generates textured and PBR 3D meshes with generative models and sparse high-resolution representations.
111 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2019–2026) · an approach in Face and shape, keyword cluster “surface · mesh · implicit”
State History Benchmarks Tried Baselines Checklist Preprints Who Export
Where it stands
Few recent reports None of the tracked benchmarks has a report from 2025–26.
21 papers in 2023–24 and 60 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 5.8% to 16.4%.
Open questions the data points to
- Papers report MeshGPT on Objaverse (COV) anywhere from 35.03 to 41.7: comparisons against it depend on whose number you take.
No longer reported by papers here: ShapeNet Chair (FID, last in 2024).
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2019 · Scan2Mesh: From Unstructured Range Scans to 3D Meshes (CVPR 2019 · 96 citations)
- 2020 · Leveraging 2D Data to Learn Textured 3D Mesh Generation (CVPR 2020 · 95 citations)
- 2021 · Deep Marching Tetrahedra: a Hybrid Representation for High-Resolution 3D Shape Synthesis (NeurIPS 2021 · 105 citations)
- 2022 · Extracting Triangular 3D Models, Materials, and Lighting From Images (CVPR 2022 · 342 citations)
- 2023 · MeshDiffusion: Score-based Generative 3D Mesh Modeling (ICLR 2023 · 30 citations)
- 2024 · MeshGPT: Generating Triangle Meshes with Decoder-Only Transformers (CVPR 2024 · 113 citations)
- 2025 · CraftsMan3D: High-fidelity Mesh Generation with 3D Native Diffusion and Interactive Geometry Refiner (CVPR 2025 · 34 citations)
- 2026 · Topology-Preserved Auto-regressive Mesh Generation in the Manner of Weaving Silk (ICLR 2026 · 1 citation)
Earliest papers here
- Scan2Mesh: From Unstructured Range Scans to 3D Meshes (CVPR 2019 · 96 citations)
- Leveraging 2D Data to Learn Textured 3D Mesh Generation (CVPR 2020 · 95 citations)
- PolyGen: An Autoregressive Generative Model of 3D Meshes (ICML 2020 · 87 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
ShapeNet Chair · FID (lower is better)
4 papers, 2022–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2022 | 23.28 | GET3D: A Generative Model of High Quality 3D Textured Shapes Learned from Images (NeurIPS 2022 · 186 citations) |
| 2024 | 9.12 | MeshXL: Neural Coordinate Field for Generative 3D Foundation Models (NeurIPS 2024 · 20 citations) |
What has been tried
75 papers here report 758 measured changes (a component added, removed or swapped, compared with a baseline): helped 72%, no clear effect 5%, hurt 20%, mixed 3%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| use learned geometric tokens instead of naive coordinate tokenization | ShapeNet Chair · COV | 27.50 -> 43.28 | Helped | MeshGPT: Generating Triangle Meshes with Decoder-Only Transformers (CVPR 2024 · 113 citations) |
| Add roughness, metalness, and normal maps to diffuse-only texture optimization | Text-driven mesh texture synthesis · visual realism | Helped | Paint-it: Text-to-Texture Synthesis via Deep Convolutional Texture Map Optimization and Physically-Based Rendering (CVPR 2024 · 37 citations) | |
| add multi-view image conditioning instead of single image | GSO subset · CD | 0.0317 -> 0.0188 | Helped | CraftsMan3D: High-fidelity Mesh Generation with 3D Native Diffusion and Interactive Geometry Refiner (CVPR 2025 · 34 citations) |
| add mesh-based training, relighting fine-tuning, and vertex offsets to TripoSR | GSO · Chamfer Distance | 0.111 -> 0.108 | Helped | SF3D: Stable Fast 3D Mesh Reconstruction with UV-unwrapping and Illumination Disentanglement (CVPR 2025 · 29 citations) |
| Increase model size from 125M to 1.3B | Objaverse · COV | 39.76 -> 42.86 | Helped | MeshXL: Neural Coordinate Field for Generative 3D Foundation Models (NeurIPS 2024 · 20 citations) |
| Use CSD instead of standard SDS | Local edits · average score | 4.06 -> 4.88 | Helped | 3D Paintbrush: Local Stylization of 3D Shapes with Cascaded Score Distillation (CVPR 2024 · 16 citations) |
| Use Gaussian Ψ instead of logistic Ψ | DTU · mean Chamfer distance | 1.98 -> 1.78 | Helped | Objects as Volumes: A Stochastic Geometry View of Opaque Solids (CVPR 2024 · 13 citations) |
| Add the diffusion prior to Lh-only optimization | APAP-BENCH 2D · k-NN GIQA (×10−2) | 4.797 -> 4.887 | Helped | As-Plausible-As-Possible: Plausibility-Aware Mesh Deformation Using 2D Diffusion Priors (CVPR 2024 · 9 citations) |
| Increase inference λ from 5 to 8 to 11 | Robot lion stylization · stylization strength | Helped | Geometry in Style: 3D Stylization via Surface Normal Deformation (CVPR 2025 · 8 citations) | |
| Add the noise-resistant decoder | Mesh extraction evaluation · CD | 2.423 -> 2.256 | Helped | MeshAnything: Artist-Created Mesh Generation with Autoregressive Transformers (ICLR 2025 · 8 citations) |
| add 2D deformation to multiview images before unprojection | GSO · appearance clarity | Helped | Fancy123: One Image to High-Quality 3D Mesh Generation via Plug-and-Play Deformation (CVPR 2025 · 6 citations) | |
| Use MLPMixer rather than Transformer backbone | UHM12k · AE mean Euclidean distance (×10−2mm) | 8.79 -> 8.36 | Helped | Locally Adaptive Neural 3D Morphable Models (CVPR 2024 · 6 citations) |
Baseline numbers that disagree
The same baseline on the same benchmark and metric, reported more than 5% apart by different papers here. Training budgets, data splits and reimplementations differ, so state which number you compare with.
| Baseline | Benchmark | Metric | Reported | Papers |
|---|---|---|---|---|
| MeshGPT | Objaverse | COV | 35.03 – 41.7 | 2 papers, e.g. ICLR 2025 and ICLR 2025 |
| MeshGPT | Objaverse | 1-NNA | 63.86 – 67.3 | 2 papers, e.g. ICLR 2025 and ICLR 2025 |
What reviewers will expect
Datasets and baselines named in the 27 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- Objaverse · 9 of 27 papers
- ShapeNet · 3 of 27 papers
- Google Scanned Objects · 3 of 27 papers
- Objaverse-XL · 3 of 27 papers
- DTU · 2 of 27 papers
- GSO · 2 of 27 papers
- OmniObject3D · 2 of 27 papers
Baselines
- InstantMesh · 5 of 27 papers
- TripoSR · 3 of 27 papers
- TEXTure · 3 of 27 papers
- PolyGen · 3 of 27 papers
- MeshGPT · 3 of 27 papers
- NeRFactor · 2 of 27 papers
- One-2-3-45 · 2 of 27 papers
- Magic123 · 2 of 27 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 81 papers; from OpenAlex and the papers' affiliations).
- Adobe · 4 papers
- University of Montreal · 3 papers
- South China University of Technology · 3 papers
- The University of Hong Kong · 3 papers
- Meta · 3 papers
- Tencent · 3 papers
- Southern University of Science and Technology · 2 papers
- The Hong Kong Polytechnic University · 2 papers
- Shanghai Jiao Tong University · 2 papers
- Johns Hopkins University · 2 papers
Use it
Download BibTeX of 37 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Neural implicit surface reconstruction (131 papers)
- Learned mesh and geometric representations (73 papers)
- Implicit neural representation design (52 papers)
- Pixel-aligned implicit human and hand reconstruction (39 papers)
Elsewhere in the topic
- Gaussian and implicit head avatars (115 papers)
- Primitive-based shape abstraction (99 papers)
- Single-view 3D reconstruction (96 papers)
- Face reflectance and relightable rendering (85 papers)
- 3D morphable face model regression (80 papers)
- Full-body avatar reconstruction (78 papers)
- Articulated and non-rigid shape learning (75 papers)
- 3D shape generative modeling (74 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).