AI Paper Atlas

Research map / 3D vision / NeRF and radiance fields

F2-NeRF: Fast Neural Radiance Field Training With Free Camera Trajectories

Peng Wang, Yuan Liu, Zhaoxi Chen, Lingjie Liu, Ziwei Liu, Taku Komura, Christian Theobalt, Wenping Wang

CVPR 2023 · 3D vision › NeRF and radiance fields · approach: Dynamic and editable NeRF · 123 citations (OpenAlex)

Read the paper

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 changedWhereMeasuredEffect
Use perspective warping instead of inverse-sphere warping (ablation)F²-NeRF with exponential sampling · Free dataset, PillarPSNR: 27.86 -> 28.65Improved
Use perspective sampling instead of exponential sampling (ablation)F²-NeRF with perspective warping · Free dataset, PillarPSNR: 28.65 -> 28.76Improved
Use inverse-sphere warping instead of no warping (ablation)F²-NeRF with disparity sampling · Free dataset, PillarPSNR: 26.81 -> 27.09Improved
Replace disparity sampling with exponential sampling (ablation)F²-NeRF with inverse-sphere warping · Free dataset, PillarPSNR: 27.09 -> 27.86Improved
Increase training from 20k to 50k steps (ablation)F²-NeRF · Free datasetPSNR: 26.32 -> 26.85Improved
Increase hash table size from 2^19 to 2^20 (ablation)F²-NeRF trained for 50k steps · Free datasetPSNR: 26.85 -> 27.19Improved
Train for 50k instead of 20k steps (ablation)Instant-NGP · Free datasetPSNR: 24.43 -> 25.07Improved
Use perspective warping instead of NDC warping (ablation)MLP-based NeRF · LLFFPSNR: 26.31 -> 26.29No clear effect
Use perspective warping instead of no warping (ablation)MLP-based NeRF · Lego view extrapolationPSNR: 37.15 -> 37.63; 30.35 -> 30.19; 27.03 -> 27.35Mixed
Use a single global hash table with multiple hash functions rather than separate hash tables per octree node (ablation)F²-NeRF · Free datasetrendering qualityImproved
Use disparity sampling rather than exponential sampling (ablation)F²-NeRF with inverse-sphere warping · Free dataset, PillarPSNR: 27.86 -> 27.09Hurt
Use perspective warping and sampling rather than inverse-sphere warping and exponential sampling (ablation)F²-NeRF · Free dataset, PillarSSIM: 0.751 -> 0.798Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy