Research map / 3D vision / NeRF and radiance fields
Cascaded and Generalizable Neural Radiance Fields for Fast View Synthesis
Phong Nguyen-Ha, Lam Huynh, Esa Rahtu, Jiří Matas, Janne Heikkilä
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 convolutional neural renderer on a coarse branch instead of random-ray rendering (ablation) | CG-NeRF · Forward-Facing | PSNR: 19.15 -> 20.59 | Improved |
| add coarse reconstruction loss L_coarse (ablation) | CG-NeRF · Forward-Facing | PSNR: 20.59 -> 22.19 | Improved |
| add hinge GAN loss with dual discriminator (ablation) | CG-NeRF · Forward-Facing | PSNR: 22.19 -> 23.93 | Improved |
| add hinge GAN loss with dual discriminator (ablation) | CG-NeRF · Forward-Facing | LPIPS: 0.25 -> 0.21 | Improved |
| add L_coarse (ablation) | CG-NeRF · Forward-Facing | SSIM: 0.76 -> 0.80 | Improved |
| add proposed neural renderer to PointNeRF | PointNeRF · Forward-Facing 800x800 | rendering time (s): 10 -> 2.2 | Improved |
| add proposed neural renderer to PointNeRF | PointNeRF · Forward-Facing 800x800 | PSNR: 26.35 -> 27.26 | Improved |
| add proposed neural renderer to Instant-NGP | Instant-NGP · Forward-Facing 4K | rendering time (s): 19 -> 6 | Improved |
| add proposed neural renderer to Instant-NGP | Instant-NGP · Forward-Facing 4K | PSNR: 26.45 -> 26.78 | Improved |
| increase reference images from 4 to 7 (ablation) | CG-NeRF · Tanks & Temples | PSNR: 18.82 -> 21.28 | Improved |
| increase reference images from 7 to 10 (ablation) | CG-NeRF · Tanks & Temples | PSNR: 21.28 -> 21.03 | Hurt |
| fine-tune generalizable model per scene for 15 minutes | CG-NeRF · Synthetic-NeRF | PSNR: 25.01 -> 34.12 | Improved |
Closest papers by meaning
- Neural Radiance Fields From Sparse RGB-D Images for High-Quality View Synthesis (TPAMI 2022 · similarity 0.93)
- Efficient View Synthesis with Neural Radiance Distribution Field (ICCV 2023 · similarity 0.93)
- NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis (ECCV 2020 · similarity 0.93)
- R2L: Distilling Neural Radiance Field to Neural Light Field for Efficient Novel View Synthesis (ECCV 2022 · similarity 0.93)
- Learning Robust Generalizable Radiance Field with Visibility and Feature Augmented Point Representation (ICLR 2024 · similarity 0.92)
- NeRF-VPT: Learning Novel View Representations with Neural Radiance Fields via View Prompt Tuning (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.
