I'm interested in using similar generative adversarial networks to reduce artifacting in video streams. For example, highly compressed streams tend to show blocking artifacts on dark scenes, gradients, and static that could be smoothed in the decoder.

I haven't actually done much about it yet, but I'm interested.