Literature Review#

Source: Awesome-Neural-Rendering

Best Tutorials#

  • Fourier Feature Networks and Neural Volume Rendering - Video

Shows how to learning a signal based on Fourier features. Does time series, 2D spatial and 2D volume.


Basics#


Most Interesting#


Best Code#



  • Plenoxels: Radiance Fields without Neural Networks - Fridovich-Keil & Yu et al (2022) - Project

They used a sparse voxel grid along with a trilinear interpolation field to fill in the missing data. Superfast convergence

pixelNeRF: Neural Radiance Fields from One or Few Images - Yu et al (2021) - Project

They use convolutions to get some global structure. Apparently it works on very few samples.

RegNeRF: Regularizing Neural Radiance Fields for View Synthesis from Sparse Inputs - Project

They learn from very few samples. They regularize with a normalizing flow.

NeRF-VAE: A Geometry Aware 3D Scene Generative Model - Paper | Talk

Learned Initializations for Optimizing Coordinate-Based Neural Representation - Project

Talks about meta-learning which can be used for some more advanced problems for faster convergence.


Spatio-Temporals#

  • Neural Radiance Flow for 4D View Synthesis and Video Processing - Du et al (2021) - Project

  • Neural Scene Flow Fields for Space-Time View Synthesis of Dynamic Scenes - Li et al (2021) - Project


Generative Models#

Normalizing Flows#

Neural Radiance Flow for 4D View Synthesis and Video Processing - Du et al (2021) - ICCV - Project


Random Fourier Features#

  • Random Feature Expansions for Deep Gaussian Processes - Cutajar et al - Thesis

  • On the Error of Random Fourier Features - Paper |

Fourier Operators#