I’m illustrating my book, In a Dog’s World, chapter by chapter for posting to Deep Sky Anchor, and it’s weird how much the workflow has changed since I did this for Discovery of the Wanderlust in October.
Back then, it was easiest to work directly with the image generators.
Now, it’s easiest to have Fable read the book and write out a full set of matching prompts. Then when Gemini/GPT1.5 give me trouble, I complain to Fable, and it reworks the whole document of prompts.
Basically, Fable’s easier to collaborate with, even though it doesn’t draw.
Gemini does really lovely illustrations, but it’s such a pain to work with…
It drew a nice picture with some extra background characters. So, I asked it to find all the characters who weren’t anthro cats or dogs wearing clothes and fix them. I failed to say IN THIS PICTURE…
So, Gemini asked for internet access, presumably to turn all characters everywhere into anthro cats and dogs.
I mean, part of me is in favor of that… but it’s also just a ridiculous response.
Okay, there is no way I could have illustrated a whole 26 chapter book in one afternoon the last time I did this. It would have easily been a 3-5 day job.
Using Fable to manage my interactions with the other AIs saves a whole lot of time.