Research map / 3D vision / NeRF and radiance fields
NerfAcc: Efficient Sampling Accelerates NeRFs
Ruilong Li, Hang Gao, Matthew Tancik, Angjoo Kanazawa
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 occupancy grid sampling (NerfAcc) | NeRF · NeRF-Synthetic | training speedup: 20x speedup, 31.00 -> 31.55 PSNR | Improved |
| add occupancy grid sampling (NerfAcc) | TensoRF · Tanks&Temples | training time (min): 18.3 -> 12.6 min, 1.5x | Improved |
| add occupancy grid sampling (NerfAcc) | TensoRF · Tanks&Temples | PSNR: 28.13 -> 28.10 | No clear effect |
| add occupancy grid sampling (NerfAcc) | T-NeRF · D-NeRF | PSNR: 28.78 -> 32.22, 20x speedup | Improved |
| add occupancy grid sampling (NerfAcc) | BARF · NeRF-Synthetic (perturbed cameras) | PSNR: 28.83 -> 30.11, 586 -> 130 min | Improved |
| add occupancy grid sampling (NerfAcc) | BARF · NeRF-Synthetic (perturbed cameras) | camera rotation error: 0.19 -> 0.07 | Improved |
| add occupancy grid sampling (NerfAcc) | TiNeuVox · D-NeRF | training time (min): 11.8 -> 4.2 min, 2.8x | Improved |
| replace proposal estimator with occupancy grid (NerfAcc) | K-Planes · D-NeRF | training time (min): 63.9 -> 38.8 min, 1.6x | Improved |
| use proposal network instead of occupancy grid (ablation) | Instant-NGP · Mip-NeRF 360 | PSNR: 26.41 -> 27.58 | Improved |
| use proposal network instead of occupancy grid (ablation) | Instant-NGP · NeRF-Synthetic | PSNR: 32.55 -> 31.40 | Hurt |
| add occupancy grid on top of proposal network (combined sampling) (ablation) | Instant-NGP · NeRF-Synthetic | PSNR: 31.40 -> 32.35 dB, 5.2 -> 4.3 min | Improved |
| add time-conditioned proposal network instead of occupancy grid (ablation) | TiNeuVox · HyperNeRF | PSNR: 24.19 -> 24.26 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- BARF — BARF: Bundle-Adjusting Neural Radiance Fields (ICCV 2021) · better on 4 of 4 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
- Efficient NeRF Optimization - Not All Samples Remain Equally Hard (ECCV 2024 · similarity 0.91)
- Rad-NeRF: Ray-decoupled Training of Neural Radiance Field (NeurIPS 2024 · similarity 0.89)
- A General Implicit Framework for Fast NeRF Composition and Rendering (AAAI 2024 · similarity 0.88)
- GL-NeRF: Gauss-Laguerre Quadrature Enables Training-Free NeRF Acceleration (NeurIPS 2024 · similarity 0.88)
- L0-Sampler: An L0 Model Guided Volume Sampling for NeRF (CVPR 2024 · similarity 0.88)
- NeRF On-the-go: Exploiting Uncertainty for Distractor-free NeRFs in the Wild (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.
