This is literally the first AI image processing project I've seen posted here that actually shows high resolution images.
Every single other one I've seen has a bunch of tiny low res thumbnails on a github page that serve to completely obscure any potential artifacts or issues with the system.
(I could clone their code and run it, but that's not the output that any of the discussion they've prompted is operating on, and that's kind of the point of hacker news).
So thanks for doing the bare minimum for an image processing project, finally.
Most of the projects you're thinking of probably never even generated high-resolution images. Until recently, the actual output layer for most of these systems would be something like a 256x256 array of pixels.
1. What use cases are there for a photo processing algorithm that only spits out tiny thumbnails?
2. If it can output higher resolution images, why are all of the examples tiny thumbnails? You can hide a lot of otherwise obvious flaws with a tiny thumbnail.
Because of computing power issues - training a good model with a significantly higher resolution becomes a lot more expensive. If you're doing a proof of concept or analyzing algorithms, you stick to lower resolutions; at larger resolutions the algorithm (and its effects) are the same, but you just need ten or hundred or thousand times more hardware and/or time.
Every single other one I've seen has a bunch of tiny low res thumbnails on a github page that serve to completely obscure any potential artifacts or issues with the system.
(I could clone their code and run it, but that's not the output that any of the discussion they've prompted is operating on, and that's kind of the point of hacker news).
So thanks for doing the bare minimum for an image processing project, finally.