revisiting my inital thoughts on stable diffusion's dataset

back in september [1] i made some experiments with stable diffusion, and today i came across a paper that did the same thing, only Properly, so i've written a bit of a follow-up on it. don't really feel like converting it to gemtext at the moment, so check it out on my website [2] ^v^

./20220907.gmi

https://solflo.neocities.org/artlog/marrow05.html