Research map / 3D vision / NeRF and radiance fields
FreeNeRF: Improving Few-Shot Neural Rendering With Free Frequency Regularization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 frequency curriculum and occlusion regularization to NeRF | NeRF on Blender · Blender (8 views) | PSNR: 13.931 -> 24.259 | Helped |
| Add frequency curriculum and occlusion regularization to mip-NeRF | mip-NeRF on DTU · DTU object (3-view) | PSNR: 9.10 -> 19.92 | Helped |
| Add frequency curriculum and occlusion regularization to mip-NeRF | mip-NeRF on LLFF · LLFF (3-view) | PSNR: 16.11 -> 19.63 | Helped |
| Add occlusion regularization (ablation) | FreeNeRF on DTU-3 · DTU-3 | PSNR: 17.40 -> 19.81 | Helped |
| Add occlusion regularization (ablation) | FreeNeRF on LLFF-3 · LLFF-3 | PSNR: 19.16 -> 19.70 | Helped |
| Replace occlusion regularization with distortion loss (ablation) | FreeNeRF on DTU-3 · DTU-3 | PSNR: 19.81 -> 15.14 | Hurt |
| Replace occlusion regularization with distortion loss (ablation) | FreeNeRF on LLFF-3 · LLFF-3 | PSNR: 19.70 -> 19.08 | Hurt |
| Extend the frequency curriculum from the 50%-schedule to the 90%-schedule (ablation) | FreeNeRF on DTU, 9 views · DTU (9-view) | object PSNR: 25.38 -> 25.59 | Helped |
| Extend the frequency curriculum from the 50%-schedule to the 90%-schedule (ablation) | FreeNeRF on DTU, 9 views · DTU (9-view) | object LPIPS: 0.096 -> 0.117 | Hurt |
| Increase occlusion range M from 10 to 20 (ablation) | FreeNeRF on DTU-3 · DTU-3 | PSNR: 19.81 -> 13.76 | Hurt |
| Adjust the near bound instead of applying occlusion regularization (ablation) | FreeNeRF on DTU-3 · DTU-3 | PSNR: 17.40 -> 17.40 | No clear effect |
| Add occlusion regularization to mip-NeRF | mip-NeRF on DTU-3 · DTU-3 | depth error: 131.97 -> 14.89 | Helped |
Closest papers by meaning
- Spatial Annealing for Efficient Few-shot Neural Rendering (AAAI 2025 · similarity 0.92)
- FrugalNeRF: Fast Convergence for Extreme Few-shot Novel View Synthesis without Learned Priors (CVPR 2025 · similarity 0.91)
- Few-Shot Neural Radiance Fields under Unconstrained Illumination (AAAI 2024 · similarity 0.90)
- NeRF On-the-go: Exploiting Uncertainty for Distractor-free NeRFs in the Wild (CVPR 2024 · similarity 0.89)
- Rad-NeRF: Ray-decoupled Training of Neural Radiance Field (NeurIPS 2024 · similarity 0.89)
- Is Vanilla MLP in Neural Radiance Field Enough for Few-shot View Synthesis? (CVPR 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.