The power loom approach to illustration

The other day I was adding some notes to my Mirabilis plot summary. I have a good 15,000 words now, as well as the full scripts for the first half of the Spring book. 

Could I even write it today? Certainly it was far easier to contemplate when Leo and I were on a roll. We had little luck with publishers (variously uninterested or unable to get the books out) and the obstacle to going it alone was the cost of all the artwork.

These days Leo is busy with some interesting but very different projects. I could complete the story as a novel (if talk is cheap, prose isn't that much more costly) but it really feels like it should be a comic. What about using AI? Not a popular choice, I know, but it's not like I'd be doing a human artist out of a job, seeing as I can't afford a human artist in the first place.

The AI isn't there yet, as these samples show. I'd need the characters to look like themselves and remain consistent from panel to panel. Also to have the right number of fingers (human artists have the edge there) and not to come out with two left hands (a mistake the robots have picked up from people).

First above is an image by Nightcafe. I don't mind the style but it hasn't got Jack and Estelle right. Still, it did better than Bing Image Creator, which first got Jack mixed up with Harry Potter and then with John Constantine. It did have a crack at lettering, though. Next is Wombo Dream's attempt at a whole page, which looks like a comic you might find in the Dreaming. Trying to redeem itself, it next goes too manga and lovey-dovey for my tastes, and gives us two Jacks into the bargain. Back to Bing (below) for what could be from a future season of Doctor Who. And at the bottom another Bing image that's either channeling Barry Windsor-Smith or trying to look like an actual Edwardian drawing. Or I guess it could be a Steeleye Span album.

What I haven't tried yet is Midjourney, the crème de la crème of generative art models. With AI advancing as it is, it might be ready within a year or two to take my thumbnail layouts and descriptions and turn those into something halfway decent. Then the only question is whether I still have those characters and stories in me.