Research map / 3D vision / Face and shape
DMesh++: An Efficient Differentiable Mesh for Complex Shapes
Sanghyun Son, Matheus Gadelha, Yang Zhou, Matthew Fisher, Zexiang Xu, Yi-Ling Qiao, Ming C. Lin, Yi Zhou
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 |
|---|---|---|---|
| replace DMesh's weighted Delaunay tessellation with the Minimum-Ball condition | DMesh · 2D font reconstruction | time (sec): 30.39 -> 11.33 | Improved |
| replace DMesh's weighted Delaunay tessellation with the Minimum-Ball condition | DMesh · 2D font reconstruction | Chamfer Distance (x1e-6): 1.97 -> 1.82 | Improved |
| replace DMesh's weighted Delaunay tessellation with the Minimum-Ball condition | DMesh · 3D point cloud reconstruction (50 models, oriented) | Chamfer Distance (x1e-3): 0.154 -> 0.033 | Improved |
| replace DMesh's weighted Delaunay tessellation with the Minimum-Ball condition | DMesh · 3D point cloud reconstruction (50 models, oriented) | time (sec): 1147 -> 282 | Improved |
| replace DMesh's weighted Delaunay tessellation with the Minimum-Ball condition | DMesh · 3D multi-view reconstruction | Chamfer Distance (x1e-3): 0.697 -> 0.342 | Improved |
| replace DMesh's weighted Delaunay tessellation with the Minimum-Ball condition | DMesh · 3D multi-view reconstruction | time (sec): 772 -> 205 | Improved |
| replace DMesh's weighted Delaunay tessellation with the Minimum-Ball condition | DMesh · 3D tessellation face probability computation | speed: up to 16 times faster in 2D and 32 times faster in 3D | Improved |
| replace DMesh's weighted Delaunay tessellation with the Minimum-Ball condition | DMesh · tessellation GPU memory | GPU memory: up to 96% less in 2D and 75% less in 3D | Improved |
| use oriented instead of unoriented point clouds as input | DMesh++ · 3D point cloud reconstruction (50 models) | Chamfer Distance (x1e-3): 0.034 -> 0.033 | No clear effect |
| use DMesh++ instead of PSR/VoroMesh/PoNQ on open-surface models | point cloud reconstruction · 3D point cloud reconstruction (500 Thingi10K models, open surfaces) | Chamfer Distance (x1e-4): 7.86 -> 0.37 | Improved |
| use DMesh++ instead of PoNQ on closed-surface models | point cloud reconstruction · 3D point cloud reconstruction (500 Thingi10K models, closed surfaces) | Chamfer Distance (x1e-4): 2.44 -> 0.37 | Improved |
| use DMesh++ instead of oriented-input DMesh on F1 and runtime tradeoff versus PSR | point cloud reconstruction · 3D point cloud reconstruction (50 models, oriented) | time (sec): 4 -> 282 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- DMesh — DMesh: A Differentiable Mesh Representation (NeurIPS 2024) · better on 9 of 9 shared results
- PoNQ — PoNQ: a Neural QEM-based Mesh Representation (CVPR 2024) · better on 4 of 5 shared results
- VoroMesh — VoroMesh: Learning Watertight Surface Meshes with Voronoi Diagrams (ICCV 2023) · better on 2 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
- DMesh: A Differentiable Mesh Representation (NeurIPS 2024 · similarity 0.93)
- QUADify: Extracting Meshes with Pixel-level Details and Materials from Images (CVPR 2024 · similarity 0.86)
- Shape Abstraction via Marching Differentiable Support Functions (CVPR 2025 · similarity 0.85)
- MeshCoder: LLM-Powered Structured Mesh Code Generation from Point Clouds (NeurIPS 2025 · similarity 0.85)
- Ghost on the Shell: An Expressive Representation of General 3D Shapes (ICLR 2024 · similarity 0.85)
- PivotMesh: Generic 3D Mesh Generation via Pivot Vertices Guidance (ICLR 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.
