August is for dabbling with AI nonsense!
AI artwork can look cool. It can also be awful and ethically crummy.
My kids start school in a little over a week now and, let me tell you, I’m sort of looking forward to it. For a summer that’s gone as quickly as it has, it’s also felt long. I haven’t been able to do as much work as I’d like to, for the obvious reasons of my kids being home every day, on top of all the stuff that happened in July with the flooring in the house, the kids’ birthday party and whatever else.
Which reminds me, I still owe a few folks some books via snail mail. I didn’t forget, don’t worry.
A few weeks ago I found myself in the beta for the AI image generator, MidJourney. It was all the rage for a while and everything I saw looked terrible. The faces were something out of a nightmare. While it could do some landscapes just fine, outside of producing visually interesting images that couldn’t break beyond the surrealist barrier, it was nothing to write home about. Yesterday I fell down a hole of seeing how much it has changed and the results are… a lot better, although still not there yet.
I spent a lot of time fiddling with it and anything space or starship related, it sort of failed miserably at. So, for me? I’m not sure I have a ton of use for it yet. The other part of that is there’s a human element involved. I have a very human cover designer I contract and he does good work. When I wanted an illustration for Intergalactic Bastard, I found one on a marketplace and paid him a few hundred dollars and he gave me exactly what I wanted. I picked him according to his art style and experience and the result was perfect.
There’s obviously an allure of spending $10 a month as opposed to hundreds of dollars, but as someone who’s an artist, I understand getting paid is important. There are already AI writing apps out there, and there are plenty of authors willing to jump on those bandwagons to become the “AI writing experts,” going as far as using them in some of their work to set some sort of example.
While I have dabbled in the AI audiobooks, I see that as a different thing altogether. Google built a learning algorithm on top of a text-to-speech program to make it sound less robotic. The work is already complete, and this is just a different format being created. Creating art or writing from scratch? That’s some perilous territory. In my mind, that means I’ll be happy to dabble in the art stuff and create some images here and there, but don’t see it as a replacement for actual, human artists.
Anyway, the above image is what I was able to generate for Zed/Zara from Trystero. Below is a portrait attempt, which came out sorta fine and then using some of Photoshop’s AI filters to touch them up, giving me an entirely too young looking character. Oh well.