Research map / 3D vision / Face and shape
Neural implicit surface reconstruction
Reconstructs surfaces from multi-view images with signed or unsigned distance fields and volume rendering.
131 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Face and shape, keyword cluster “surface · mesh · implicit”
State History Tried Baselines Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
46 papers in 2023–24 and 41 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 12.7% to 11.2%.
Open questions the data points to
- Papers report NeuS on DTU (Chamfer Distance) anywhere from 0.77 to 0.87: comparisons against it depend on whose number you take.
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
- 2016 · Single Image Object Modeling Based on BRDF and R-Surfaces Learning (CVPR 2016 · 3 citations)
- 2018 · SobolevFusion: 3D Reconstruction of Scenes Undergoing Free Non-Rigid Motion (CVPR 2018 · 115 citations)
- 2019 · DeepSDF: Learning Continuous Signed Distance Functions for Shape Representation (CVPR 2019 · 3,528 citations)
- 2020 · DIST: Rendering Deep Implicit Signed Distance Function With Differentiable Sphere Tracing (CVPR 2020 · 263 citations)
- 2021 · Neural Geometric Level of Detail: Real-Time Rendering With Implicit 3D Shapes (CVPR 2021 · 439 citations)
- 2022 · Improving Neural Implicit Surfaces Geometry With Patch Warping (CVPR 2022 · 121 citations)
- 2023 · Neuralangelo: High-Fidelity Neural Surface Reconstruction (CVPR 2023 · 431 citations)
- 2024 · NeuSurf: On-Surface Priors for Neural Surface Reconstruction from Sparse Input Views (AAAI 2024 · 26 citations)
- 2025 · Details Enhancement in Unsigned Distance Field Learning for High-fidelity 3D Surface Reconstruction (AAAI 2025 · 12 citations)
- 2026 · Single-Stage fMRI-to-3D Reconstruction via Viewpoint-Aware Embedding and Hierarchical Guidance (AAAI 2026 · 1 citation)
Earliest papers here
- Single Image Object Modeling Based on BRDF and R-Surfaces Learning (CVPR 2016 · 3 citations)
- SobolevFusion: 3D Reconstruction of Scenes Undergoing Free Non-Rigid Motion (CVPR 2018 · 115 citations)
- DeepSDF: Learning Continuous Signed Distance Functions for Shape Representation (CVPR 2019 · 3,528 citations)
What has been tried
103 papers here report 999 measured changes (a component added, removed or swapped, compared with a baseline): helped 73%, no clear effect 5%, hurt 19%, mixed 2%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add global and local geometry losses | DTU little-overlap · Mean Chamfer Distance | 2.46 -> 1.35 | Helped | NeuSurf: On-Surface Priors for Neural Surface Reconstruction from Sparse Input Views (AAAI 2024 · 26 citations) |
| use multi-resolution hash encoding instead of positional encoding | DiLiGenT-MV (5-object average), SDM-UniPS normals · L2 Chamfer distance [mm] | 0.228 -> 0.194 | Helped | SuperNormal: Neural Surface Reconstruction via Multi-View Normal Integration (CVPR 2024 · 15 citations) |
| remove the Eikonal loss | DeepFashion · NC | 0.020 -> 0.033 | Hurt | DUDF: Differentiable Unsigned Distance Fields with Hyperbolic Scaling (CVPR 2024 · 14 citations) |
| add physical loss | ScanNet++ · Object stability SR (%) | 25.28 -> 68.96 | Helped | PhyRecon: Physically Plausible Neural Scene Reconstruction (NeurIPS 2024 · 13 citations) |
| Apply absolute value to the SIREN output instead of using an unconditioned output | Stonewall, Stanford 3D Scene · CD-mean | 2.69 -> 5.27 | Hurt | Details Enhancement in Unsigned Distance Field Learning for High-fidelity 3D Surface Reconstruction (AAAI 2025 · 12 citations) |
| Add differentiable-rendering optimization | Straight Hair, 2mm/20° · F1 | 83.4 -> 86.2 | Helped | Dr.Hair: Reconstructing Scalp-Connected Hair Strands without Pre-Training via Differentiable Rendering of Line Segments (CVPR 2024 · 10 citations) |
| add segmentation loss to the translation module | Real PX⇒CBCT · Dice (%) | 66.08 -> 67.00 | Helped | NeBLa: Neural Beer-Lambert for 3D Reconstruction of Oral Structures from Panoramic Radiographs (AAAI 2024 · 9 citations) |
| add Lder | Nandi the Bull · CD-GEP (*105) | 0.238 -> 0.228 | Helped | Unsigned Orthogonal Distance Fields: An Accurate Neural Implicit Representation for Diverse 3D Shapes (CVPR 2024 · 9 citations) |
| add the appearance and geometry optimization stage | mip-NeRF 360 outdoor mean · PSNR | 18.1 -> 23.2 | Helped | LTM: Lightweight Textured Mesh Extraction and Refinement of Large Unbounded Scenes for Efficient Storage and Real-time Rendering (CVPR 2024 · 6 citations) |
| Add point-based visibility testing | Occluded regions · surface normal accuracy | Helped | Differentiable Point-based Inverse Rendering (CVPR 2024 · 5 citations) | |
| Add a diffusion prior | 9 real-world scenes · Acc. [cm] ↓ | 0.99 -> 0.88 | Helped | MorpheuS: Neural Dynamic 360deg Surface Reconstruction from Monocular RGB-D Video (CVPR 2024 · 5 citations) |
| Add sparsity regularization and basis BRDF merge and removal control | Synthetic scenes · basis BRDF interpretability | Helped | Differentiable Inverse Rendering with Interpretable Basis BRDFs (CVPR 2025 · 4 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 |
|---|---|---|---|---|
| NeuS | DTU | Chamfer Distance | 0.77 – 0.87 | 2 papers, e.g. CVPR 2023 and CVPR 2023 |
| VolSDF* | DTU | Chamfer distance (Mean) | 0.86 – 0.94 | 2 papers, e.g. CVPR 2023 and CVPR 2023 |
| NeuS | DTU | Chamfer distance (Mean) | 0.78 – 0.84 | 2 papers, e.g. CVPR 2023 and CVPR 2023 |
What reviewers will expect
Datasets and baselines named in the 24 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- DTU · 3 of 24 papers
- ScanNet · 3 of 24 papers
- Replica · 3 of 24 papers
- BlendedMVS · 2 of 24 papers
- Tanks and Temples · 2 of 24 papers
- Objaverse · 2 of 24 papers
Baselines
- COLMAP · 4 of 24 papers
- VolSDF · 3 of 24 papers
- CAP-UDF · 2 of 24 papers
- PANDORA · 2 of 24 papers
- NeISF · 2 of 24 papers
- NeuS · 2 of 24 papers
- NeuRIS · 2 of 24 papers
- Neuralangelo · 2 of 24 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 87 papers; from OpenAlex and the papers' affiliations).
- Zhejiang University · 6 papers
- Tsinghua University · 4 papers
- Meta · 3 papers
- Chinese Academy of Sciences · 3 papers
- Nanyang Technological University · 3 papers
- Alibaba Group · 3 papers
- Max Planck Institute for Informatics · 3 papers
- EPFL · 3 papers
- University of Chinese Academy of Sciences · 2 papers
- Simon Fraser University · 2 papers
Use it
Download BibTeX of 40 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
- Textured mesh generation (111 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).