Six days, a couple all-nighters, and apparently several thousand images later (...I wouldn't believe it, but that's what my logs say), and the thing that's impressing me most right now is how versatile this thing is.

I'll spend hours going down a rabbit hole trying to refine a particular look, but then when I finally back out and try something else, it's impressive to see it come up with something *entirely* different.

It's all the same pipeline, running off the same model. There is no switching out the backend between rendering illustrations of vogons in a rocking chair, to photographs of cats, to engravings of muppet-themed bank notes.

It's one engine that can replicate any style that's sufficiently well-represented in its training set of billions of images. 🤯

Sign in to participate in the conversation

On the internet, everyone knows you're a cat — and that's totally okay.