Research map / 3D vision / NeRF and radiance fields
Strata-NeRF : Neural Radiance Fields for Stratified Scenes
Ankit Dhiman, R Srinath, Harsh Rangwani, Rishubh Parihar, Lokesh R Boregowda, Srinath Sridhar, R Venkatesh Babu
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 |
|---|---|---|---|
| train a single mip360 on all levels instead of level-wise models | mip-NeRF 360 · Dragon In Pyramid Level 1 | PSNR: 29.8560 -> 24.3876 | Hurt |
| add VQ-VAE latent generator and latent router to mip-NeRF 360 | mip-NeRF 360 · Cube-Sphere-Monkey Total | PSNR: 24.0909 -> 26.2778 | Improved |
| add Strata-NeRF to mip360 on real-estate scenes with more levels | mip-NeRF 360 · RealEstate10K 7 Rutledge Ave | PSNR: 19.621 -> 25.040 | Improved |
| add Strata-NeRF to mip360 on six-level scene | mip-NeRF 360 · Synthetic six-level simple scene | PSNR: 15.35 -> 23.54 | Improved |
| use VQ-VAE discrete latent instead of VAE (ablation) | Strata-NeRF · RealEstate10K average | PSNR: 20.99 -> 24.75 | Improved |
| add latent router instead of direct concatenation of generated embedding (ablation) | Strata-NeRF · Synthetic average | PSNR: 27.41 -> 28.25 | Improved |
| use both routers in latent router instead of only one (ablation) | Strata-NeRF · Synthetic average | PSNR: 26.04 -> 28.25 | Improved |
| share codebook across levels instead of independent codebooks (ablation) | Strata-NeRF · RealEstate10K 7 Rutledge Ave | PSNR: 23.53 -> 25.04 | Improved |
| choose codebook size 1024 versus 512 or 4096 (ablation) | Strata-NeRF · Buddhist Temple Total | PSNR: 23.3769 (4096) -> 26.9343 (1024) | Improved |
| use Strata-NeRF instead of mip360 | mip-NeRF 360 · Spanish Colonial Retreat | training time: 5h 30m -> 6h 20m | Hurt |
Closest papers by meaning
- NeRFReN: Neural Radiance Fields With Reflections (CVPR 2022 · similarity 0.92)
- AligNeRF: High-Fidelity Neural Radiance Fields via Alignment-Aware Training (CVPR 2023 · similarity 0.92)
- Aug-NeRF: Training Stronger Neural Radiance Fields With Triple-Level Physically-Grounded Augmentations (CVPR 2022 · similarity 0.92)
- Mask-Based Modeling for Neural Radiance Fields (ICLR 2024 · similarity 0.90)
- NeRF On-the-go: Exploiting Uncertainty for Distractor-free NeRFs in the Wild (CVPR 2024 · similarity 0.90)
- ProvNeRF: Modeling per Point Provenance in NeRFs as a Stochastic Field (NeurIPS 2024 · similarity 0.90)
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.
