Building off the original NeRF image rendering technique published 5 months ago, this new iteration is able to take a set of images from very complex scenes to create super smooth moving pictures without the previous fog and consistency issues. This level of execution and understanding of how lighting changes with viewpoint has promising real world applications, such as photography for tourist attractions like the Trevi Fountain.
"NeRF in the Wild - Neural Radiance Fields for Unconstrained Photo Collections" by Ricardo Martin-Brualla, Noha Radwan,
Mehdi S. M. Sajjadi, Jonathan T. Barron, Alexey Dosovitskiy and Daniel Duckworth:
https://nerf-w.github.io/
