Research map / 3D vision / NeRF and radiance fields
Single-view and streaming radiance fields
Builds conditioned, recolorable or streaming radiance fields from limited or sequential observations.
49 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2021–2026) · an approach in NeRF and radiance fields, keyword cluster “radiance fields · fields nerf · nerfs”
State History Benchmarks Tried Checklist Preprints Who Export
Where it stands
Quiet None of the tracked benchmarks has a report from 2025–26.
31 papers in 2023–24 and 5 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 11.7% to 5.6%.
No longer reported by papers here: LLFF (PSNR, last in 2024).
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
- 2021 · Sharf: Shape-conditioned Radiance Fields from a Single View (ICML 2021 · 44 citations)
- 2022 · Plenoxels: Radiance Fields Without Neural Networks (CVPR 2022 · 1,375 citations)
- 2023 · K-Planes: Explicit Radiance Fields in Space, Time, and Appearance (CVPR 2023 · 545 citations)
- 2024 · GARField: Group Anything with Radiance Fields (CVPR 2024 · 53 citations)
- 2025 · ShowMak3r: Compositional TV Show Reconstruction (CVPR 2025 · 2 citations)
- 2026 · Radiance Meshes for Volumetric Reconstruction (CVPR 2026 · 1 citation)
Earliest papers here
- Sharf: Shape-conditioned Radiance Fields from a Single View (ICML 2021 · 44 citations)
- TöRF: Time-of-Flight Radiance Fields for Dynamic Scene View Synthesis (NeurIPS 2021 · 10 citations)
- Generative Occupancy Fields for 3D Surface-Aware Image Synthesis (NeurIPS 2021 · 4 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
LLFF · PSNR (higher is better)
4 papers, 2022–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2022 | 26.41 | Unsupervised Multi-View Object Segmentation Using Radiance Field Propagation (NeurIPS 2022 · 6 citations) |
| 2023 | 26.92 | K-Planes: Explicit Radiance Fields in Space, Time, and Appearance (CVPR 2023 · 545 citations) |
| 2024 | 27.16 | Entangled View-Epipolar Information Aggregation for Generalizable Neural Radiance Fields (CVPR 2024 · 7 citations) |
What has been tried
45 papers here report 461 measured changes (a component added, removed or swapped, compared with a baseline): helped 75%, no clear effect 4%, hurt 18%, mixed 3%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add scale-conditioning | ramen, teatime, keyboard, bouquet, and living room · mIOU | 64.1 -> 74.9; 67.7 -> 64.9; 57.6 -> 23.2; 49.8 -> 34.4; 49.7 -> 59.6 | Mixed | GARField: Group Anything with Radiance Fields (CVPR 2024 · 53 citations) |
| Add RGB conditioning | Touch estimation (novel views) test set · FID | 34.31 -> 28.97 | Helped | Tactile-Augmented Radiance Fields (CVPR 2024 · 28 citations) |
| add geometry features to nearest matching | RGB-D stylization · stylization diversity and content preservation | Helped | Geometry Transfer for Stylizing Radiance Fields (CVPR 2024 · 10 citations) | |
| Add along-epipolar perception to the view transformer | LLFF low-resolution · PSNR | 25.03 -> 25.48 | Helped | Entangled View-Epipolar Information Aggregation for Generalizable Neural Radiance Fields (CVPR 2024 · 7 citations) |
| add ACA, SVA spatial and detail modules, and CAF | DTU, 3-view · PSNR | 27.73 -> 29.36 | Helped | Geometry-aware Reconstruction and Fusion-refined Rendering for Generalizable Neural Radiance Fields (CVPR 2024 · 7 citations) |
| Add reliable-region reconstruction loss Lrel | Realistic Synthetic 360°, 4-view · PSNR | 17.19 -> 17.79 | Helped | Learning with Unreliability: Fast Few-shot Voxel Radiance Fields with Relative Geometric Consistency (CVPR 2024 · 7 citations) |
| Remove the forward step | Real dataset · PSNR | 27.73 -> 27.64 | Hurt | OmniLocalRF: Omnidirectional Local Radiance Fields from Dynamic Videos (CVPR 2024 · 7 citations) |
| Add VGG feature distillation loss | 3D scene style transfer · stylization quality | Helped | FPRF: Feed-Forward Photorealistic Style Transfer of Large-Scale 3D Neural Radiance Fields (AAAI 2024 · 6 citations) | |
| Remove coarse style transfer | Plenoptic Video · Stylization quality | Hurt | S-DyRF: Reference-Based Stylized Radiance Fields for Dynamic Scenes (CVPR 2024 · 6 citations) | |
| Train only the last color MLP layer instead of the full color MLP | NeRF Synthetic, LLFF, and Mip NeRF 360 · PSNR | 28.25 -> 29.09; 24.86 -> 24.99; 24.31 -> 25.35 | Helped | IReNe: Instant Recoloring of Neural Radiance Fields (CVPR 2024 · 5 citations) |
| Set mask overlapping rate to 0.25 rather than 0.125 | Synthetic Dataset · PSNR | 31.93 -> 33.61 | Helped | SCINeRF: Neural Radiance Fields from a Snapshot Compressive Image (CVPR 2024 · 5 citations) |
| add face-fitting refinement | Face reconstruction · PSNR | 21.52 -> 24.34 | Helped | ShowMak3r: Compositional TV Show Reconstruction (CVPR 2025 · 2 citations) |
What reviewers will expect
Datasets and baselines named in the 18 papers of this approach from 2024–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- LLFF · 6 of 18 papers
- NeRF Synthetic · 3 of 18 papers
- DTU · 2 of 18 papers
- LERF dataset · 2 of 18 papers
- Mip-NeRF 360 · 2 of 18 papers
Baselines
- 3DGS · 4 of 18 papers
- PixelNeRF · 3 of 18 papers
- MVSNeRF · 3 of 18 papers
- NeRF · 3 of 18 papers
- IBRNet · 2 of 18 papers
- NeuRay · 2 of 18 papers
- GNT · 2 of 18 papers
- MatchNeRF · 2 of 18 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 36 papers; from OpenAlex and the papers' affiliations).
- Pohang University of Science and Technology · 3 papers
- Seoul National University · 3 papers
- The Chinese University of Hong Kong · 3 papers
- Meta · 3 papers
- Korea Advanced Institute of Science and Technology · 3 papers
- ETH Zurich · 2 papers
- University of California, Los Angeles · 2 papers
- University of California, Berkeley · 2 papers
- University of Maryland, College Park · 2 papers
- National Taiwan University · 2 papers
Use it
Download BibTeX of 38 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
- Sparse-input NeRF regularization (106 papers)
- Dynamic and editable NeRF (88 papers)
- NeRF for scene decomposition and understanding (30 papers)
Elsewhere in the topic
- Inverse rendering with neural fields (110 papers)
- Few-shot neural rendering architectures (53 papers)
- 3D-aware generative radiance fields (20 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).