AI video models can now do impressive style-transfer like this to convert a video between different styles. For example, converting a real-life video into anime, or "pixar" style. However doing this is very time-consuming (about 20 seconds to create 1 second of video on an RTX4090).

When will it be possible to convert videos (with quality at least this good) in real time using consumer hardware. By consumer hardware, I mean a PC (or other hardware) that can be bought by anyone for less than $10k (2023 dollars).