Despite increasing computing power, high quality media still presents a storage problem. If the raw file is compressed to piss poor quality with an algorithm, the same algorithm could be used to faithfully (or with a margin of error outside our sensorial capabilities) live upscale the media file back while you are playing it. So you could have a full 8k video library resting on 720p files, lossless audio coming from 192kbps files, and stuff like that, a quality good enough to embed for piss poor connections and processing power, but that can be locally live upscaled by the algorithm on better devices.
ps: Upscaling media is already real, it’s just bad at the moment. Funny to see people believing it won’t get better. How easily are you still telling AI videos apart from the real thing without the aid of professional tools? Just a year ago it wasn’t so difficult.
AI upscale is already real, and if compression behaves always the same way it’s easier to recreate what was lost.
No
Once detail is removed it can never be restored faithfully. Ai can generate details and an individual may not detect much difference but in direct comparison it will be different.
If you somehow include the details as data all combined in a compressed format you save space but the result is not a lower res, but a lossless compressed native res in which case upscaling is irrelevant.
Let’s try to explain compression with a mathematical example:
You have the numbers 10, 35, 56 and 59. You form an average out of those values, which is 40. Now you give that average number to me. You tell me it’s made up of 4 numbers and ask me what those numbers were. I can form a guess, based on other examples you send me before and told me afterwards what numbers were in there. But I will at most randomly guess the right values by chance.
Compression is not different from this example, were the mathematical numbers represent colors.