Nvidia developed a radically different way to compress video calls

Good idea. Technically this is an extremely lossy compression as only the most important data from specific features is transmitted. I still would differentiate it from mpeg, because it is probably not generic for any pixel image.

A client could verify the result of the morphological operations with the real image to check the result. Computationally expensive, but Nvidia probably has a solution for that.

I still hate video calls though.