Research map / 3D vision / 3D generation
UniTEX: Universal High Fidelity Generative Texturing for 3D Shapes
Yixun Liang, Kunming Luo, Xiao Chen, Rui Chen, Hongyu Yan, Weiyu Li, Jiarui Liu, Fei-Peng Tian, Ping Tan
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 Texture Function supervision (volumetric) instead of surface-only supervision (ablation) | Large Texturing Model · full-model texturing | PSNR: 25.81 -> 27.01 | Improved |
| add Texture Function supervision (ablation) | Large Texturing Model · full-model texturing | PSNR_uv: 20.31 -> 20.99 | Improved |
| drop 50% of tokens during MV generation training (drop training) (ablation) | Flux MV generation · artist-created mesh | training memory (GB): 69.2GB -> 53.6GB | Improved |
| add drop training (ablation) | Flux MV generation · artist-created mesh | training step time (s/it): 38.76s/it -> 21.50s/it | Improved |
| add drop training (ablation) | Flux MV generation · artist-created mesh | FID_CLIP: 17.49 -> 16.03 | No clear effect |
| add drop training (ablation) | Flux MV generation · artist-created mesh | LPIPS: 0.087 -> 0.090 | No clear effect |
| use 3D LTM refinement instead of UV-based refinement | UniTEX second stage · artist-created mesh | PSNR_uv: 20.47 -> 23.01 | Improved |
| use UniTEX instead of Hunyuan3D-Paint | texturing pipeline · artist-created mesh | FID_CLIP: 20.38 -> 16.03 | Improved |
| use UniTEX instead of Hunyuan3D-Paint | texturing pipeline · artist-created mesh | LPIPS: 0.091 -> 0.090 | No clear effect |
| use UniTEX instead of Hunyuan3D-Paint | texturing pipeline · generative mesh | user preference (%): 21.36% -> 65.91% | Improved |
| use native 3D single-stage TexGaussians instead of 2D-prior two-stage | texturing pipeline · generative mesh | CLIPscore: 0.807 -> 0.681 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Paint3D — Paint3D: Paint Anything 3D with Lighting-Less Texture Diffusion Models (CVPR 2024) · better on 3 of 3 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
- Paint3D: Paint Anything 3D with Lighting-Less Texture Diffusion Models (CVPR 2024 · similarity 0.88)
- TUVF: Learning Generalizable Texture UV Radiance Fields (ICLR 2024 · similarity 0.88)
- GOATex: Geometry & Occlusion-Aware Texturing (NeurIPS 2025 · similarity 0.88)
- TriTex: Learning Texture from a Single Mesh via Triplane Semantic Features (CVPR 2025 · similarity 0.88)
- Wukong's 72 Transformations: High-fidelity Textured 3D Morphing via Flow Models (NeurIPS 2025 · similarity 0.87)
- MeshGen: Generating PBR Textured Mesh with Render-Enhanced Auto-Encoder and Generative Data Augmentation (CVPR 2025 · 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.
