Computer Vision News - September 2020

He points out that they are not the first group to try to represent scenes as a continuous function of the network, but they have found a novel way to make it work well for complex and photo-realistic content . They have used computer vision to represent shapes as continuous functions and got that to represent detailed and photo-realistic content, which is more exciting from a graphics perspective. Now, someof these techniques that have been building up in the computer vision community can be used to produce very high-resolution pictures. The whole team were thrilled to see the work gain a keen following after it went up on arXiv. “People are already doing all kinds of cool follow-up projects!” Pratul exclaims. “We are all so excited to see the different ideas people have and how they are building on our original NeRF concept.” 3 NeRF: Pratul Srinivasan 35 Best of ECCV 2020