Okay, so I've been trying out Anima lately so I wanted to write this down as a memorial to all the beans that got burned in the process.

Let's start with ILL. It's like a painter with a solid grounding in aesthetics.

For example, if you tell it "a girl, red dress," it'll give you a pretty decent-looking girl in red.

But when you ask for more, like "she's carrying a basket with a wolf's head in it," it might go "what the hell are you talking about? Ugly!" and just decide on its own to turn it into a basket with a sandwich instead.

Looks nice, sure, but it's not what I asked for.

Anima, on the other hand, is more like a drawing assistant. Whatever you tell it to do, it does. It doesn't act according to its own aesthetic preferences.

So it might give you a really rough, sloppy girl in a red dress, then shrug and go, "huh? You never said you wanted it detailed."

But when you tell it to make it detailed, it actually delivers something detailed. Tell it to draw the basket + wolf head, and it draws it.

So the prompt ends up having a much bigger impact on the output.

For instance, one of my OCs has olive-green hair (though I'm honestly not 100% sure what color olive green actually is).

When I made her with ILL, it automatically picked a more natural-looking green. But with Anima, when I saw the actual olive green, my first reaction was, "this color is so ugly, the hell are you doing!?" My second reaction was, "wait... that's actually what olive green looks like 🤡"

So I had to start going back and re-examining my OCs' tags to make them more precise.

Annoying, honestly. But there's something really satisfying about slowly learning and making progress.

(#1 Olive green by Anima, #2 Olive green by ILL)