Research map / 3D vision / NeRF and radiance fields
Baking Neural Radiance Fields for Real-Time View Synthesis
Peter Hedman, Pratul P. Srinivasan, Ben Mildenhall, Jonathan T. Barron, Paul Debevec
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 deferred rendering (view-dependence once per pixel) (ablation) | SNeRG · Synthetic 360 scenes | ms/frame: 343.6 -> 11.9 | Improved |
| add sparsity loss (ablation) | SNeRG · Synthetic 360 scenes | ms/frame: 20.0 -> 11.9 | Improved |
| add sparsity loss (ablation) | SNeRG · Synthetic 360 scenes | GPU memory (GB): 4.26 -> 1.73 | Improved |
| add sparsity loss (ablation) | SNeRG (PNG) · Synthetic 360 scenes | storage (MB): 176.0 -> 86.7 | Improved |
| add sparsity loss (ablation) | SNeRG (PNG) · Synthetic 360 scenes | PSNR: 30.22 -> 30.38 | No clear effect |
| add fine-tuning of the view-dependence network after baking (ablation) | SNeRG (PNG) · Synthetic 360 scenes | PSNR: 26.68 -> 30.38 | Improved |
| add view-dependence MLP (ablation) | SNeRG · Synthetic 360 scenes | ms/frame: 9.2 -> 11.9 | No clear effect |
| add view-dependent appearance network versus diffuse only (ablation) | JAXNeRF+ · Synthetic 360 scenes | PSNR: 27.39 -> 33.00 | Improved |
| replace per-sample view network with deferred per-pixel evaluation (ablation) | JAXNeRF+ · Synthetic 360 scenes | PSNR: 31.65 -> 30.55 | Hurt |
| compress texture atlas with H264 instead of PNG (ablation) | SNeRG · Synthetic 360 scenes | PSNR and storage: 86.7 -> 30.2 MB, 30.38 -> 29.86 PSNR | Mixed |
| lower voxel grid resolution from 1000^3 to 500^3 (ablation) | SNeRG (PNG) · Synthetic 360 scenes | FPS and PSNR: 84.06 -> 101.78 FPS, 30.38 -> 28.93 PSNR | Mixed |
Closest papers by meaning
- NeRFLight: Fast and Light Neural Radiance Fields Using a Shared Feature Grid (CVPR 2023 · similarity 0.93)
- Learning Neural Duplex Radiance Fields for Real-Time View Synthesis (CVPR 2023 · similarity 0.92)
- PyNeRF: Pyramidal Neural Radiance Fields (NeurIPS 2023 · similarity 0.92)
- Neural Radiance Fields From Sparse RGB-D Images for High-Quality View Synthesis (TPAMI 2022 · similarity 0.92)
- Efficient View Synthesis with Neural Radiance Distribution Field (ICCV 2023 · similarity 0.92)
- Ced-NeRF: A Compact and Efficient Method for Dynamic Neural Radiance Fields (AAAI 2024 · similarity 0.91)
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.
