24.9.23 note: This page was made in 2021, when image generation was basically a novelty and getting results that looked "real" seemed impossibly far off. It was made before the release of Stable Diffusion and Midjourney, before image generation became as simple as typing in a sentence on Discord.
I find this early era of text2img fascinating and inspiring, because it creates images that are wonderfully impossible and fucked up, obviously not the work of any human artist or photographer and not one that we (humans) can easily replicate.
In hindsight, I have been too optimistic about this technology; the programs that I used to make these images have since been completely abandoned because everyone's seeking something more "human", something that wears the skin of a flesh-and-blood artist. I do not share this affection.