High-fidelity 3D mesh generation from a single sketch using shape constraints
Yingbin Wu, Fubo Wang, Peng Zhao, Mingquan Zhou, Shengling Geng, Dan Zhang · 2025
Read High-fidelity 3D mesh generation from a single sketch using shape constraints by Yingbin Wu, Fubo Wang, Peng Zhao, Mingquan Zhou, Shengling Geng, Dan Zhang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-12789542.txt.
Description
Yingbin Wu and co-authors turn a single hand sketch into a detailed three-dimensional mesh, in work from twenty twenty-five. Rebuilding 3D models from photos has advanced fast with deep learning. Sketches carry far less visual detail than photos, though. That sparseness makes correct reading of shape much harder. Their network uses a compact encoder-decoder build for the task.
A stronger multi-layer design, PowerMLP, powers the core network. Shape constraints guide the build instead of a rival judge network. Those constraints keep the mesh true to real geometry in training. Tests cover both clean synthetic sketches and messy real hand drawings. Scores reach top levels on both test sets in the report. Robustness across drawing styles is a key selling point here. Designers could draft products by hand and get solid models.
Catalog record
| CREATOR | Yingbin Wu, Fubo Wang, Peng Zhao, Mingquan Zhou, Shengling Geng, Dan Zhang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 69 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |