Скачать или смотреть The Crazy World of Neural Rendering

The field of Neural Rendering has grown enormously over the past 2 years. The field has already contributed to txt-to-3d, single-shot image-to-3d, meshless 3d, and novel view synthesis with much more promise for the future. These methods have proven to be both high quality, flexible, and robust. Let's explore Neural Rendering!

Timestamps:
0:00 Intro
0:56 NeRF
2:00 NGP + mip360
3:25 SNeRG + NeRV + dNeRF
6:03 Single views
7:18 Latent Dif + Neural

Repos
torch-ngp - https://github.com/ashawkey/torch-ngp
ngp_pl - https://github.com/kwea123/ngp_pl

Academic Papers:
I ran out of characters with standard citations. You should be able to google the arXiv citation, if you have trouble find a source let me know! Many of the papers from 2022 may still be in pre-print (non-peer reviewed).

[0] arXiv:2111.11426 - nerf review
[1] arXiv:2210.00379 - nerf paper
[2] arXiv:2201.05989 - ngp paper
[3] arXiv:2111.12077 - mip360 paper
[4] arXiv:2103.14645 - snerg paper
[5] arXiv:2012.03927 - nerv paper
[6] arXiv:2011.13961 - dnerf paper
[7] arXiv:2204.00928 - sinnerf paper
[8] arXiv:2210.08936 - s^3 nerf paper
[9] arXiv:2212.03267 - nerdi paper
[10] arXiv:1901.05103 - deepSDF paper
[11] arXiv:2104.09877 - shadow nerfs paper
[12] arXiv:2211.10440 - magic3d paper
[13] arXiv:2209.14988 - dreamfusion paper
Not used but referred to
[14] arXiv:2203.12575 - neuMan paper
[15] arXiv:2203.07182 - NeILF paper

Thanks arXiv for being free!