Research map / 3D vision / Face and shape
DeepMesh: Differentiable Iso-Surface Extraction
Benoît Guillard, Edoardo Remelli, Artem Lukoianov, Pierre Yvernay, Stephan R. Richter, Timur Bagautdinov, Pierre Baqué, Pascal Fua
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 contour-matching refinement through differentiable mesh (ablation) | DeepMesh SDF · ShapeNet car SVR | Chamfer distance x1e4: 2.96 -> 2.56 | Improved |
| add contour-matching refinement through differentiable mesh (ablation) | DeepMesh SDF · ShapeNet chair SVR | Chamfer distance x1e4: 9.07 -> 8.22 | Improved |
| add differentiable-rendering silhouette refinement (ablation) | DeepMesh SDF · ShapeNet car SVR | Chamfer distance x1e4: 2.96 -> 2.73 | Improved |
| use SDF instead of occupancy field (ablation) | DeepMesh refinement · ShapeNet chair SVR | Chamfer distance x1e4: 10.35 -> 8.22 | Improved |
| add refinement through differentiable mesh (ablation) | DeepMesh occupancy · ShapeNet car SVR | Chamfer distance x1e4: 3.02 -> 2.65 | Improved |
| add refinement on real images (ablation) | DeepMesh (raw) · Pix3D Chairs | Chamfer distance x1e3: 4.850 -> 4.063 | Improved |
| use learned DeepMesh parameterization instead of hand-crafted FreeForm | FreeForm · CFD drag minimization | relative drag objective: 0.844 -> 0.675 | Improved |
| use SDF instead of occupancy network (ablation) | DeepMesh · CFD drag minimization | relative drag objective: 0.721 -> 0.675 | Improved |
| add mesh-based refinement of latent features (ablation) | ConvOccNet · Scene reconstruction | Chamfer distance: 0.954 -> 0.529 | Improved |
| add mesh-based refinement of latent features (ablation) | ConvOccNet · Scene reconstruction | IoU %: 81.67 % -> 88.74 % | Improved |
| add Chamfer loss to implicit loss in training (ablation) | DeepSDF · ShapeNet Lamps | Chamfer distance x1e4: 8.12 -> 7.59 | Improved |
| add Chamfer loss to implicit loss in training (ablation) | DeepSDF · ShapeNet Chairs | Chamfer distance x1e4: 2.75 -> 2.56 | Improved |
Later papers that compared with this method
4 later accepted papers report DeepMesh as a baseline in their results tables; 4 of them report a better number on most of the results they share with it.
- Mesh BDF: Barycentric Dominance Field for 3D Native Mesh Generation (NeurIPS 2026 · better on 6 of 6 shared results)
- MeshRipple: Structured Autoregressive Generation of Artist-Meshes (CVPR 2026 · better on 5 of 6 shared results)
- LATO: 3D Mesh Flow Matching with Structured TOpology Preserving LAtents (ICML 2026 · better on 4 of 4 shared results)
- MeshWeaver: Sparse-Voxel-Guided Surface Weaving for Autoregressive Mesh Generation (CVPR 2026 · better on 1 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Pixel2Mesh — Pixel2Mesh: Generating 3D Mesh Models from Single RGB Images (ECCV 2018) · better on 4 of 4 shared results
- DeepSDF — DeepSDF: Learning Continuous Signed Distance Functions for Shape Representation (CVPR 2019) · better on 2 of 2 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
- MeshSDF: Differentiable Iso-Surface Extraction (NeurIPS 2020 · similarity 0.92)
- Learning and Meshing From Deep Implicit Surface Networks Using an Efficient Implementation of Analytic Marching (TPAMI 2021 · similarity 0.89)
- DMesh: A Differentiable Mesh Representation (NeurIPS 2024 · similarity 0.89)
- Iso-Points: Optimizing Neural Implicit Surfaces With Hybrid Representations (CVPR 2021 · similarity 0.88)
- MIND: Material Interface Generation from UDFs for Non-Manifold Surface Reconstruction (NeurIPS 2025 · similarity 0.87)
- DUDF: Differentiable Unsigned Distance Fields with Hyperbolic Scaling (CVPR 2024 · 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.
