The team behind this breakthrough was led by Wenbo Bao from Shanghai Jiao Tong University, and included computer scientists from the University of California Merced, and Google. Together, they used the power of deep convoluted neural networks to significantly improve the quality and capability of video frame interpolation, to the point where you’d be hard-pressed to spot any artifacts.

Guys need to understand that DNN used here is not some universal interpolation algorithm, it is frequently specifically trained thing that can suddenly utterly break or not work at all and hence having lot of some band aids that require research (and can be fully absent from scientific article).

Interesting is that this guys will never promote stuff like Prodad software as it is not clickbait and they can't pay for promotion.

Neural networks field is now utterly dirty with lot of unfounded claims, falsifications and more of the same. People hope that nice financing and stupid managers will continue and it'll be enough for their time.

In some fields that are directly opposite from "universal breakthroughs" and usually use lots and lots of servers and resources we have good progress - like cloud based voice recognition that have DNN but also lot of other algorithms combined together.

https://sites.google.com/view/wenbobao/dain