Developing your own style of AI art

If you’re an artist, geek, or someone who frequents the internet at some point every day, you’ve probably heard about AI artwork. I myself have written about AI art.. A bit.

As AI art becomes more evolved and common, artists will need to distinguish themselves so they stand out awash a sea of visually interesting pieces.

AI art itself lends itself well to a “throw things against a wall and see what sticks” approach, using repeated prompting and iteration until you make something that “looks cool”. That’s fine for some, but as I’ve said on Twitter.. You’re basically creating the muzak (or “Wal-Mart safe”) version of art. Stuff you’d see at a dentist’s office, rather than in a gallery.

Much like a series of acrylic barn paintings, without innovation all you have is a hope that your prompt is more interesting than the other artists’.. Which is going to become less and less likely as more artists begin to learn the basics of AI art generation.

So, you’ve got to innovate. Don’t just stick with Midjourney and pick and choose the best outputs; work with it.

I’m lucky in that I already have an artists’ style before getting into AI art. I’ve exhibited and sold pieces for a long time, so even when I’m relying heavily on the AI, I’m still working on a core visual style that can’t be purely described to an AI all the time. That means I have to utilize a multi-step process.

Most of my major pieces that utilize AI start out with an image. In the case of “The Ghost”, I utilized the brand mark of my gaming news website Ghost Gamer News as a base.. Manipulating it a bit to ensure the proper definition to match my prompting.

I took multiple results, cut them together, retoned/recolored some aspects, and even repainted a few areas before upscaling in Topaz Gigapixel AI (which works well for this sort of thing).. But I wasn’t done yet.

This piece was designed to be more in my photography style, which is exceptionally distinct: I have very particular lighting and camera settings I use for nearly all my modern photos. Lighting was easy; the image input already encouraged that aspect properly, and I just had to re-tone for color.

Afterwards, I ended up utilizing Adobe’s Camera Raw tool to add in my high-ISO noise and do some final tweaks until I got the final work.

Sure, I could’ve just fed it a prompt and let it iterate a few hundred or thousand times until it came up with something “close” that looked cool.. There were a few options like that, in fact.

It never would’ve come out to be the same, however, if I hadn’t already developed my style, which is recognizable to fans of my work.

Let yourself be creative with an AI for art.. Both in it’s usage, and what you do with the output, too.