Research map / 3D vision / NeRF and radiance fields
PlenOctrees for Real-Time Rendering of Neural Radiance Fields
Alex Yu, Ruilong Li, Matthew Tancik, Hao Li, Ren Ng, 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 |
|---|---|---|---|
| fine-tune the octree on training images after conversion (ablation) | PlenOctree · NeRF-synthetic | PSNR: 31.02 -> 31.71 | Improved |
| fine-tune the octree on training images after conversion (ablation) | PlenOctree · Tanks and Temples | PSNR: 27.34 -> 27.99 | Improved |
| replace NeRF with NeRF-SH (SH-factorized color) | NeRF · NeRF-synthetic | PSNR: 31.69 -> 31.57 | No clear effect |
| convert to PlenOctree for rendering instead of NeRF network queries | NeRF · NeRF-synthetic | FPS: 0.045 -> 167.68 | Improved |
| project NeRF to SH by Monte Carlo sampling instead of training NeRF-SH (ablation) | NeRF to PlenOctree · NeRF-synthetic | PSNR: 31.02 -> 29.21 | Hurt |
| raise weight threshold tau_w to 0.01 (ablation) | PlenOctree · NeRF-synthetic | PSNR: 31.71 -> 31.64 | Mixed |
| add automatic bounding box scaling (ablation) | PlenOctree · NeRF-synthetic | PSNR: 30.70 -> 31.71 | Improved |
| use grid size 512 instead of 256 (ablation) | PlenOctree · NeRF-synthetic | PSNR: 29.60 -> 31.71 | Improved |
| use 16 SH basis functions instead of 9 (ablation) | PlenOctree · NeRF-synthetic | PSNR: 31.45 -> 31.71 | Improved |
| use 25 SH basis functions instead of 16 (ablation) | PlenOctree · NeRF-synthetic | PSNR: 31.71 -> 31.69 | No clear effect |
| use spherical Gaussians instead of SH-16 (ablation) | PlenOctree · NeRF-synthetic | PSNR: 31.57 -> 31.74 | No clear effect |
| stop NeRF-SH training early and fine-tune PlenOctree | NeRF-SH · NeRF-synthetic ship scene | training time: 16 hours -> 4.5 hours | Improved |
Closest papers by meaning
- Dynamic PlenOctree for Adaptive Sampling Refinement in Explicit NeRF (ICCV 2023 · similarity 0.93)
- Fourier PlenOctrees for Dynamic Radiance Field Rendering in Real-Time (CVPR 2022 · similarity 0.92)
- PyNeRF: Pyramidal Neural Radiance Fields (NeurIPS 2023 · similarity 0.91)
- Rad-NeRF: Ray-decoupled Training of Neural Radiance Field (NeurIPS 2024 · similarity 0.91)
- Ced-NeRF: A Compact and Efficient Method for Dynamic Neural Radiance Fields (AAAI 2024 · similarity 0.89)
- A General Implicit Framework for Fast NeRF Composition and Rendering (AAAI 2024 · similarity 0.89)
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.
