Research map / 3D vision / NeRF and radiance fields
LOLNerf: Learn From One Look
Daniel Rebain, Mark Matthews, Kwang Moo Yi, Dmitry Lagun, Andrea Tagliasacchi
Check 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 |
|---|---|---|---|
| Use LOLNeRF instead of π-GAN for training-image reconstruction | CelebA training-set reconstruction · CelebA training-set reconstruction, 128² | PSNR: 23.5 -> 29.1 | Improved |
| Use LOLNeRF instead of π-GAN for held-out image reconstruction | FFHQ held-out images · FFHQ held-out image reconstruction, 256² | PSNR: 21.8 -> 26.2 | Improved |
| Use LOLNeRF instead of π-GAN for high-resolution held-out reconstruction | FFHQ held-out images · FFHQ held-out image reconstruction, 512² | PSNR: 20.9 -> 25.3 | Improved |
| Use LOLNeRF instead of π-GAN for novel-view synthesis | HUMBI face images · HUMBI novel view synthesis | PSNR: 24.5 -> 27.0 | Improved |
| Use LOLNeRF rather than supervised depth prediction methods | 3DFAW · 3DFAW depth prediction | Depth Correlation: 50.18 vs 50.81; 50.18 vs 58.68 | Mixed |
| Use LOLNeRF instead of π-GAN for unconditional generation | CelebA unconditional sampling · CelebA unconditional sampling | FID: 14.7 -> 128.2 | Hurt |
| Use LOLNeRF instead of π-GAN for unconditional generation | CelebA unconditional sampling · CelebA unconditional sampling | KID: 0.39 -> 0.11 | Improved |
| Use LOLNeRF instead of π-GAN for unconditional generation | CelebA unconditional sampling · CelebA unconditional sampling | IS: 2.62 -> 2.34 | Mixed |
| Increase the training set from very small subsets to more images (ablation) | LOLNeRF · Dataset-size ablation | Novel-view geometric consistency | Improved |
| Increase the training set from very small subsets to more images (ablation) | LOLNeRF · Dataset-size ablation | Training-image reconstruction accuracy | Hurt |
| Add the hard-surface weight prior (ablation) | LOLNeRF single-view shape reconstruction · Single-view object reconstruction | Shape quality | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- GRAF — GRAF: Generative Radiance Fields for 3D-Aware Image Synthesis (NeurIPS 2020) · better on 1 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- GRF: Learning a General Radiance Field for 3D Representation and Rendering (ICCV 2021 · similarity 0.91)
- Sharf: Shape-conditioned Radiance Fields from a Single View (ICML 2021 · similarity 0.90)
- Stereo Radiance Fields (SRF): Learning View Synthesis for Sparse Views of Novel Scenes (CVPR 2021 · similarity 0.90)
- GRAF: Generative Radiance Fields for 3D-Aware Image Synthesis (NeurIPS 2020 · similarity 0.90)
- pixelNeRF: Neural Radiance Fields From One or Few Images (CVPR 2021 · similarity 0.89)
- AltNeRF: Learning Robust Neural Radiance Field via Alternating Depth-Pose Optimization (AAAI 2024 · similarity 0.88)
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.
